Dixa

Summary of Important Changes

1. Document structure and incorporated terms

• The opening text now identifies the document as “Privacy Policy” and adds a “Last updated: 25 Novembre 2021” reference.
• The policy is expressly stated to be incorporated into and subject to the Dixa Terms of Service, rather than merely referring to the “Dixa Terms.”
• The definition of the “Agreement” has been reformatted, but its substance appears unchanged.
• A large amount of navigation or website-index text has been inserted, including references to:
• Legal Hub
• Terms of Service
• Cookie Policy
• GDPR
• Third-Party Services
• Data Processing Agreement
• Data Security
• Applicant Privacy Notice
• Global Feedback Panelist and Accessibility Statements
• Early Access Terms
• Referral Program

Risk: The inserted navigation text appears potentially accidental or poorly integrated. It could create uncertainty about which linked policies or terms form part of the contractual arrangement, particularly where the document references associated procedures and other policies.

2. Definitions and scope

• The definitions of Dixa, Website, Dixa Services, and Customer Data are re-presented with updated quotation marks and apostrophes.
• No material change to the apparent scope of Customer Data or Personal Data was identified.
• The policy continues to state that Personal Data forming part of Customer Data is subject to the customer’s own privacy policy, and that data-subject requests should generally be directed to the customer.

3. Data use and disclosure

• The policy continues to describe use of information for purposes including:
• Analyzing usage trends and preferences
• Administering and optimizing the services
• Monitoring usage and traffic patterns
• Understanding customer and visitor interests, habits, and usage patterns
• Corporate, shareholder, legal, accounting, and compliance purposes
• The changes primarily correct punctuation, quotation marks, apostrophes, and formatting.
• No new express authorization to sell, share, disclose, or otherwise use Customer Data was identified.

4. AI model training

• No provision in the supplied diff expressly addresses artificial intelligence, machine learning, model training, model improvement, or use of Customer Data to train AI models.
• The general language allowing analysis of usage trends, preferences, traffic patterns, or service optimization could potentially be broad enough to support analytics or product-improvement activities, but the diff does not specifically expand that language or clearly authorize AI training.
• Customers should not assume that the absence of an AI-training clause constitutes a prohibition. The Terms of Service, Data Processing Agreement, or Data Security documentation should also be reviewed.

5. Hosting, security, and changes

• The policy continues to state that services are hosted in the United States or EU, with EU customer databases hosted in an EU data center.
• No material change was identified regarding security breaches, subcontractors, cookies, policy updates, or contact information.
• The updated contact reference now includes privacy@dixa.com.

Overall assessment

The diff appears largely editorial and structural, with no clear substantive expansion of data-use rights and no express change concerning AI training. The main practical risk is ambiguity caused by the inserted navigation and cross-reference material, together with the need to check the separate contractual documents for AI-related permissions.

Summary

The diff states only: “Added approximately 23 words to the document.” It does not include the actual added language or identify where the words were inserted.

Key Legal Changes

• Cannot be determined: The substance and legal effect of the 23-word addition cannot be assessed without the actual text.
• No confirmed changes: There is insufficient information to identify changes involving:
• Customer data ownership or permitted uses
• Data sharing with third parties
• Confidentiality or security obligations
• Intellectual property rights
• Liability, indemnification, or compliance obligations
• Termination or post-termination data handling

AI Training and Model Use

• No identifiable AI-related change: The provided diff does not show whether customer data may be used to train, fine-tune, evaluate, or improve AI models.
• It also does not indicate whether:
• Customer data is excluded from model training by default
• Customer consent is required
• De-identified or aggregated data may be used
• Inputs, outputs, prompts, or usage metadata may be retained
• Third-party AI providers may receive customer data
• Training restrictions apply after termination

Risk Assessment

A meaningful legal risk assessment is not possible from the supplied information. The actual 23-word addition is required to determine whether it expands the provider’s rights to use customer data or introduces new customer obligations.

Recommendation: Provide the exact added text, including its location in the agreement and any surrounding language.

Structured Summary of Important Changes

1. Major expansion and restructuring of the Terms

• The document is substantially expanded and now includes a detailed definitions section and provisions covering:
• Add-Ons, including Mim AI Co-Pilot, AI Quality Assurance, Collaboration Users and Advanced Insights.
• Customer Accounts, Customer Data, Content, Permitted Users and Affiliates.
• Overage, including AI usage, API calls, messaging and telecommunications usage.
• Minimum Commitments and subscription structures.
• The Agreement now expressly incorporates the applicable Order Form, Documentation, Privacy Policy and standard Data Processing Agreement.

2. AI-related products, usage and charges

• AI products are expressly identified as paid Add-Ons. This may make AI functionality subject to separate pricing, subscription and usage restrictions.
• The definition of Overage now expressly includes:
• Analysing more data with AI tools than included in the relevant plan.
• Usage exceeding quantities or thresholds in the Order Form.
• AI-related subscriptions and other products may form part of the Customer’s Minimum Commitment, which cannot generally be reduced during the Subscription Term.
• Overage charges are non-cancellable and non-refundable once incurred, and are billed at then-current rates without discounts.
• Customers should confirm whether AI features are automatically enabled, how usage is measured, and whether AI output or processing can generate unexpected charges.

3. Customer data and AI model training

• The revised terms broadly define Customer Data to include information provided by the Customer, uploaded by users, and Content supplied by Customer End-Users.
• Dixa retains no ownership of Customer Data; the Customer remains responsible for its legality, accuracy, integrity and third-party rights.
• A new data-use provision permits Dixa to:
• Aggregate, collect and analyse data relating to the provision, use and performance of the Services, including traffic and telemetry data.
• Use that data during and after the Agreement to develop and improve the Services and other Dixa offerings.
• Disclose it only in aggregated and anonymised form that does not identify the Customer or an individual.
• This clause does not expressly mention AI, machine learning, foundation models, model training, model fine-tuning, prompts or outputs. Accordingly, it is unclear whether “develop and improve” includes training or improving AI models.
• “Aggregated” and “anonymised” are not defined, and the clause states that Dixa obtains and retains rights in the resulting data. Customers should seek written confirmation that identifiable Customer Data, prompts, transcripts, recordings and AI inputs/outputs will not be used to train shared or third-party models.
• The AI/data-use permission survives termination and may conflict in scope with the general confidentiality obligation.

4. Other significant customer risks

• Dixa may modify the Services and Documentation over time.
• Subscriptions generally renew automatically for 12-month terms unless notice is given at least three months before expiry.
• A 5% annual compound fee indexation applies automatically.
• Dixa may suspend access for breach or non-payment.
• Liability is capped at Fees paid in the preceding six months, with broad exclusions for data loss, consequential loss and third-party services.
• The Customer provides a broad indemnity for claims arising from its use of the Services and Customer Data.

Summary of Important Changes

1. New Terms of Service structure and effective date

• The Terms are now identified as “Service – 28 May 2025”, replacing the previous “EAT”/“Last updated” presentation.
• The revised agreement applies when accepted through access or use of the Services, authorization by an agent or end-user, or an Order Form.
• The agreement may therefore bind a company through actual use, even if no individual signature is mentioned.

2. New and expanded AI-related services

The revised definitions expressly introduce AI products, including:

• Mimic AI Co-Pilot
• AI Quality Assurance
• AI-based data analysis and AI-agent usage

The definition of “Minimum Commitment” now expressly includes usage-based subscriptions, such as conversations handled by AI agents. “Overage” also includes analyzing more data with AI tools than the purchased plan allows.

Risk: AI usage may create additional, non-refundable charges, and the customer’s minimum commitment cannot generally be reduced during the subscription term.

3. Customer data and AI model use

The revised confidentiality section gives Dixa the right to:

• Aggregate, collect, and analyze data and information relating to the provision, use, and performance of the Services, including traffic and telemetry data;
• Use that information during and after the agreement to develop and improve the Services and other Dixa offerings; and
• Disclose it only in aggregated and anonymized form that does not identify the Customer or any individual.

Dixa also states that it obtains and retains rights in this aggregated and anonymized data.

AI-training risk: The clause does not expressly mention “training AI models,” but “develop and improve” the Services and other offerings is broad enough potentially to include model training or tuning. The provision is not limited to the customer’s own account, service term, or a particular AI feature. It also does not expressly state whether conversation content, prompts, outputs, or derived data may be used, nor provide an opt-out.

The text does state that data must not permit identification of an individual, but anonymization standards and re-identification safeguards are not defined.

4. Customer data ownership and responsibility

• The customer retains ownership of Customer Data.
• The customer remains responsible for its legality, accuracy, integrity, and third-party rights.
• The standard DPA applies unless replaced by a separate DPA.
• Use of sensitive personal data is prohibited unless specifically agreed in the DPA.

5. Commercial and operational changes

• Automatic 5% annual compounded fee indexation applies to subscription fees.
• Dixa may immediately update list and overage pricing through the Platform.
• Subscriptions generally cannot be cancelled or downgraded during the current term.
• The agreement renews automatically for successive 12-month terms unless notice is given at least three months before expiry.
• Dixa may suspend access for violations or non-payment.

6. Liability and data-loss exposure

• Dixa’s aggregate liability is capped at fees paid during the preceding six months.
• Dixa’s remedy for Customer Data loss is limited to commercially reasonable restoration efforts from the latest backup.
• Loss or corruption of data is excluded from certain damages claims.

Summary

Scope of the Diff

The supplied diff does not include the actual contractual language. It only states:

> “Added approximately 9,125 words to the document”

Because the added text is not provided, it is not possible to determine what terms changed or whether existing provisions were replaced, deleted, or materially modified.

AI Training and Customer Data

No conclusion can be reached about whether the agreement now permits or restricts the use of customer data to train AI models.

In particular, the available diff does not reveal whether the new language:

• Allows the provider to use customer content, prompts, outputs, personal data, or usage data to train, fine-tune, test, or improve AI models;
• Limits AI training to de-identified, anonymized, aggregated, or “service-generated” data;
• Requires customer consent or provides an opt-out mechanism;
• Permits use of data for model training by affiliates, contractors, subprocessors, or third-party model providers;
• Grants the provider ownership or broad license rights over customer data or AI outputs;
• Requires deletion or segregation of customer data from training datasets;
• Addresses retention periods, confidentiality, security, or cross-border transfers in connection with AI training; or
• Provides assurances that customer data will not be used to train models.

Potential Risks Requiring Review

The added language should be reviewed for provisions that:

1. Expand permitted data use beyond providing the contracted services.

2. Define customer data broadly, including prompts, outputs, metadata, telemetry, feedback, or derived data.

3. Grant perpetual, worldwide, irrevocable, royalty-free licenses to use customer content.

4. Permit disclosure to third parties or use by affiliates and subprocessors.

5. Allow model training without express consent or an opt-out.

6. Make de-identification the provider’s only safeguard, without explaining the standard or re-identification protections.

7. Limit the provider’s liability for unauthorized AI training, privacy violations, or misuse of data.

8. Conflict with confidentiality, data-protection, retention, or deletion obligations elsewhere in the agreement.

Information Needed

To perform a meaningful legal analysis, the complete redline or the approximately 9,125 words of added text—and, preferably, the corresponding deleted and replacement language—must be provided.

Executive Summary

The diff appears to be primarily a reformatting and document-structure update, with limited substantive changes to privacy terms. The most notable change is that the document is now labeled a “Privacy Notice” and includes navigation links to other Legal Hub documents and programs. No express provision authorizing the use of Customer Data to train AI models appears in the supplied diff.

Important Changes

1. Document title and surrounding Legal Hub content

• The former heading, “Privacy Policy (“Policy”)”, is replaced or supplemented with:
• “Privacy Notice”
• “Last updated: 25 Novembre 2021”
• Links to the Terms of Service, Cookie Policy, GDPR information, Data Processing Agreement, Data Security information, Applicant Privacy Notice, and other programs.
• This may improve navigation but could create uncertainty about which linked documents govern particular processing activities.
• The date remains 25 November 2021, which may be significant if this is presented as a current notice.

2. Incorporation into the Terms of Service

• The text is revised to state that the Privacy Policy/Notice is incorporated into and subject to the Dixa Terms of Service.
• The definition of “Agreement” is reformatted from bracketed text, with no clear substantive change.
• Because the Privacy Notice is expressly incorporated into the Terms of Service, changes to either document may affect the contractual privacy framework.

3. Definitions

• Definitions of Dixa, Website, Dixa Services, and Customer Data are restyled, mainly by replacing typographic quotation marks with standard quotation marks.
• No apparent expansion or narrowing of “Customer Data” is shown in the supplied diff.
• The text continues to distinguish Personal Data included in Customer Data and states that such data is subject to the customer’s own privacy policy.

4. Data use and disclosure

The diff retains or restates existing purposes including:

• Analyzing usage trends and preferences.
• Administering, optimizing, and monitoring the Dixa Services.
• Tracking user movements within the Services.
• Gathering demographic information.
• Identifying shareholders and complying with legal or corporate obligations.
• Sharing information with affiliates, third parties, or service providers where permitted.

These passages could allow broad analytics and service-improvement activities, but the diff does not clearly add a new category of processing.

AI Model Training

• No language expressly permitting Dixa to use Customer Data, Personal Data, prompts, conversations, or other customer content to train, fine-tune, or improve artificial-intelligence or machine-learning models is visible in this diff.
• The general references to analyzing usage, optimizing services, and understanding customer interests could potentially cover product analytics, but they do not expressly authorize AI training.
• Customers should confirm whether AI-training rights appear in the Terms of Service, Data Processing Agreement, product-specific terms, or a separate AI policy.

Risk Assessment

• Low apparent substantive change in the supplied excerpts.
• Moderate interpretive risk from incorporation into the Terms of Service and broad service-improvement language.
• The redline appears mechanically corrupted in places; the complete revised document should be reviewed before relying on this comparison.

Summary of Important Changes

1. New and expanded AI-related products and charges

• “Add-On” now expressly includes Mim AI Co-Pilot, AI Quality Assurance, and Advanced Insights.
• “Minimum Commitment” expressly includes usage-based services, including conversations handled by AI agents.
• “Overage” now covers:
• Data analyzed by AI tools or services beyond the purchased allowance;
• Usage exceeding limits for AI agents and other usage-based features;
• Additional agents associated with paid Add-Ons.
• AI and other Add-On usage may therefore create additional, non-cancellable charges, with overage billed at then-current rates and potentially without discounts.
• Minimum commitments cannot generally be reduced during the subscription term, even if actual usage is lower.

2. Customer data ownership and responsibility

• The Customer retains ownership of “Customer Data,” which is broadly defined to include:
• Data provided by the Customer;
• Data uploaded by users;
• Content provided by the Customer’s end-users.
• The Customer bears sole responsibility for the legality, accuracy, integrity, and third-party rights relating to Customer Data.
• The Customer must pay royalties, fines, and settlements arising from third-party rights violations in Customer Data, without seeking contribution from Dixa.

3. AI model training and use of Customer Data

• The new terms do not expressly state that Dixa may use identifiable Customer Data or customer conversations to train general-purpose AI models.
• However, Dixa receives a broad right to aggregate, collect, and analyze data relating to the provision, use, and performance of the Services, including traffic and telemetry data.
• Dixa may use such data during and after the agreement to “develop and improve the Services and other Dixa offerings.” This language could potentially include AI development or model training, particularly where data has been aggregated and anonymized.
• Dixa may disclose the resulting data only in aggregated and anonymized form that does not identify the Customer or an individual.
• The clause does not explain:
• Whether conversation content, prompts, outputs, or feedback may be used;
• What anonymization standard applies;
• Whether the Customer can opt out;
• Whether data is used to train shared or customer-specific models;
• How re-identification risks are addressed.
• Customers handling confidential, regulated, or personal data should seek clarification or a written restriction on AI training and secondary use.

4. Data protection and international processing

• Dixa’s standard Data Processing Agreement applies unless a separate DPA takes precedence.
• Use of special-category personal data is prohibited unless specifically agreed in the DPA.
• Services and subprocessors may operate outside the EU/EEA, subject to an approved transfer mechanism such as Standard Contractual Clauses.

5. Commercial and operational risks

• Fees automatically increase by 5% annually on a compounded basis, without separate notice.
• Dixa may immediately change list and overage prices by posting them in the Platform.
• Subscriptions automatically renew for successive 12-month periods unless notice is given at least three months before expiry.
• Dixa may modify the Services and Documentation, and may suspend access for breaches or payment issues.
• Liability is capped at fees paid during the preceding six months, and data-loss remedies are limited primarily to restoration from Dixa’s latest backup.

Diff Analysis Summary

Scope of the Diff

The diff states only:

> “Added approximately 9,125 words to the document”

No text of the added provisions—or the prior agreement—is provided. Accordingly, the specific legal, commercial, and data-use implications cannot be reliably analyzed.

Customer Data and AI Training

The available diff does not identify whether customer data may be used to train, fine-tune, evaluate, or improve AI models.

The following issues therefore remain unresolved:

• Whether customer data, prompts, inputs, outputs, or personal information may be used for AI training.
• Whether use is limited to providing the contracted services or also permits product development, analytics, benchmarking, or model improvement.
• Whether the provider may use data in aggregated, de-identified, pseudonymized, or identifiable form.
• Whether customer data may be retained after termination for training or other purposes.
• Whether the customer can opt out of AI training or require deletion of data used for that purpose.
• Whether third-party AI providers or subprocessors may access or use the data.
• Whether the provider makes any commitments concerning confidentiality, security, data segregation, or ownership of model outputs.
• Whether the agreement allocates responsibility for personal data, confidential information, intellectual property, or regulatory compliance used in AI systems.

Other Legal Risks

Because the added language is not included, it is not possible to assess changes concerning:

• Liability caps, indemnities, warranties, or disclaimers.
• Intellectual property ownership and licensing.
• Confidentiality and permitted disclosures.
• Data protection and international transfers.
• Security obligations and breach notification.
• Suspension, termination, and post-termination data deletion.
• Service levels, audit rights, or unilateral amendment rights.
• Governing law and dispute resolution.

Required Information

Please provide the actual 9,125 words added to the agreement, together with the original language where provisions were replaced or deleted. A meaningful comparison should focus especially on terms such as “customer data,” “inputs,” “outputs,” “model training,” “machine learning,” “improve,” “develop,” “aggregate,” “de-identify,” “retain,” “subprocessor,” and “service data.”

Summary

The diff only states: “Added approximately 23 words to the document.” It does not include the actual added language or identify where the change appears.

Legal and Commercial Impact

• The specific legal effect cannot be assessed without the text of the 23 added words.
• It is not possible to determine whether the addition changes:
• Customer data ownership or usage rights
• Confidentiality or security obligations
• Service-provider permissions
• Liability, indemnification, or compliance responsibilities
• Automatic renewal, termination, or other commercial terms

AI Training and Data Use

• The provided diff does not reveal whether customer data may be used to train, fine-tune, evaluate, or improve AI models.
• No conclusion can be drawn about whether:
• Customer content is used for model training;
• Data is anonymized, aggregated, or retained;
• Human reviewers or third-party AI providers may access the data;
• The customer can opt out of AI training; or
• Customer data is excluded from training by default.

Required Information

Please provide the actual 23-word addition, including any surrounding sentence or section heading. Without the added wording, a reliable legal-risk analysis is not possible.

Summary of Important Changes

1. New or Expanded AI-Related Terms

• AI products are expressly defined as paid Add-Ons. The definition of “Add-On” now specifically includes products such as Mim AI Co-Pilot and AI Quality Assurance.
• AI usage may affect minimum commitments and billing. “Minimum Commitment” now expressly includes usage-based subscriptions, including conversations handled by AI agents.
• AI analysis can generate overage charges. “Overage” now includes analysing more data with AI tools or services than is included in the applicable plan. Excess usage is charged at then-current rates, is non-cancellable and non-refundable, and may be billed monthly.
• No express right to use identifiable Customer Data to train AI models appears in the new text. However, the confidentiality clause permits Dixa to aggregate, collect and analyse data relating to the provision, use and performance of the Services, including traffic and telemetry data, and to use it during and after the Agreement to develop and improve the Services and other Dixa offerings. This could potentially include improving AI features or models.
• Dixa states that this data must be handled in a way that does not permit identification of an individual and may only be disclosed in aggregated and anonymised form that does not identify the Customer or an individual. Dixa also obtains rights in that aggregated and anonymised data.
• The provision contains no express customer approval right, opt-out, retention limit, or explicit prohibition on using anonymised data for AI training. Customers seeking to restrict AI/model training should request clarification or a written amendment.

2. Customer Data and Privacy

• Customer retains ownership of Customer Data and remains responsible for its legality, accuracy and third-party rights.
• The Customer is designated as data controller, with Dixa’s standard DPA incorporated unless a separate DPA applies.
• Use of special-category/sensitive personal data is prohibited unless specifically agreed in the DPA.
• Services and service providers may operate outside the EU, subject to an approved transfer mechanism such as Standard Contractual Clauses.

3. Commercial and Commitment Risks

• Subscriptions automatically renew for successive 12-month terms unless at least three months’ notice is given.
• A 5% annual compound indexation applies automatically to subscription fees without separate notice.
• Minimum commitments cannot generally be reduced during the term, even if actual usage falls below the commitment.
• Dixa may change list and overage prices immediately by posting them in the Platform.
• Downgrades and cancellations are generally unavailable during the current term, and fees are broadly non-refundable.

4. Liability and Operational Changes

• Dixa may suspend accounts for breach or non-payment.
• Dixa’s aggregate liability is capped at fees paid during the preceding six months, while indirect losses—including data loss—are excluded.
• The customer provides a broad indemnity for claims arising from its use of the Services.
• Dixa may amend the Terms with 30 days’ notice, subject to limited exceptions, and may modify the Services and Documentation.

Summary

The provided diff only states that approximately 9,125 words were added. It does not include the actual added language or identify which provisions changed.

Key Legal Changes

• Cannot be determined from the information provided.
• The scope, effect, and legal significance of the additions cannot be assessed without the underlying text.
• It is not possible to determine whether the additions modify:
• Customer obligations
• Provider rights
• Fees or payment terms
• Confidentiality
• Data ownership
• Security obligations
• Liability or indemnification
• Termination rights
• Governing law or dispute resolution
• Use of subcontractors or third-party services

AI Training and Customer Data

• No specific AI-training language is included in the supplied diff.
• Therefore, it cannot be determined whether the new text:
• Permits the provider to use customer data to train, fine-tune, validate, or improve AI models
• Limits training use to de-identified, aggregated, or anonymized data
• Requires customer consent or provides an opt-out
• Allows human review of customer data for model development
• Grants the provider ownership or broad usage rights over inputs, outputs, or derived data
• Applies different rules to personal data, confidential information, or regulated data
• Allows retention of data after termination for AI-training purposes

Risk Assessment

The addition of 9,125 words could materially change the agreement, but no reliable risk assessment is possible without the actual additions. In particular, any provisions concerning AI, data usage, confidentiality, intellectual property, or model improvement should be reviewed closely.

Please provide the full redline or the actual added text for a substantive comparison.

Summary

The provided diff states only that approximately 23 words were added, but does not include the actual wording of those additions.

Legal and Commercial Impact

• No substantive legal changes can be identified without the text of the added language.
• It is not possible to determine whether the additions change:
• Customer obligations or rights
• Liability, indemnification, warranties, or termination rights
• Confidentiality or data-protection commitments
• Use, ownership, licensing, or disclosure of customer data
• Service levels, pricing, or other commercial terms

AI Model Training and Customer Data

• The available diff contains no text addressing AI, machine learning, model training, automated processing, or use of customer data.
• Accordingly, it is not possible to determine whether customer data may be:
• Used to train or improve AI models
• Combined with other customers’ data
• Retained after termination
• De-identified or anonymized for model development
• Shared with AI or technology providers

Required for Further Review

Please provide the actual 23-word addition, together with any surrounding language if available. The specific wording is necessary to assess whether it creates new rights to use customer data or introduces other legal risks.

Summary of Important Changes and Risks

1. New or expanded AI-related products and billing

• “Add-On” is newly defined to include AI products such as Mim AI Co-Pilot, AI Quality Assurance, and Advanced Insights.
• “Minimum Commitment” now expressly includes usage-based services, including conversations handled by AI agents. The committed volume cannot generally be reduced during the Subscription Term, even if actual usage is lower.
• “Overage” now expressly includes:
• Data analysed by AI tools or services beyond the purchased allowance;
• Usage of AI agents beyond contractual limits; and
• Other usage exceeding quantitative limits in the Order Form.
• AI-related overage is chargeable at Dixa’s then-current rates, is generally non-cancellable and non-refundable, and may be billed without a discount.

Risk: Customers may incur unexpected AI usage charges, particularly where AI processing is metered by conversations, data volume, agents, or other usage measures. The applicable pricing and thresholds may be located in an Order Form, Platform, or separate price list.

2. Customer data and AI model training

The revised terms include a new or materially expanded data-use right:

> Dixa may aggregate, collect, and analyse data and other information relating to the provision, use, and performance of the Services, including traffic and telemetric information.

Dixa may use such information during and after the Agreement to:

• Develop and improve the Services and other Dixa offerings; and
• Disclose it in aggregated and anonymised form.

The clause states that the data will not permit identification of an individual and that disclosures will not identify the Customer or any individual.

Important limitation: The clause does not expressly say that Customer Content, conversations, prompts, transcripts, or other Customer Data will never be used to train general-purpose AI models. “Develop and improve” is broad and could potentially encompass AI model development or training, depending on how Dixa interprets and implements the provision.

The clause also does not expressly provide:

• An opt-out from AI training or service improvement;
• A requirement that data be irreversibly anonymised before internal use;
• A defined retention period;
• Restrictions on use by AI vendors or subprocessors; or
• A clear distinction between operational telemetry and substantive customer content.

Customers should seek written clarification, particularly for customer conversations, personal data, confidential information, prompts, outputs, and AI-quality-assurance data.

3. Data protection and international transfers

• The Customer is expressly treated as the data controller, with Dixa’s standard Data Processing Agreement incorporated unless a separate DPA applies.
• Customers are generally prohibited from processing special-category personal data unless specifically agreed in the DPA.
• Services may be performed outside the EU, subject to an approved transfer mechanism such as Standard Contractual Clauses.

Risk: Customers remain responsible for lawful collection and use of data submitted to AI and other Services, including obtaining required notices and consents.

4. Commercial and operational changes

• Automatic 5% annual compounded fee indexation applies to subscription fees.
• Minimum commitments cannot generally be reduced during the term.
• Dixa may change Services and Documentation over time.
• Dixa may change Agreement terms with generally 30 days’ notice, with continued use constituting acceptance.
• Downgrading may result in loss of data or functionality, for which the Customer is responsible.

5. Liability and data-loss exposure

• Dixa’s aggregate liability is capped at fees paid in the preceding six months.
• Dixa’s remedy for lost or damaged Customer Data is limited to commercially reasonable restoration from the latest backup.
• Indirect losses, including data loss or corruption, are broadly excluded.

Summary of Important Changes

1. Entire Terms of Service Added

The diff replaces a brief introductory section with a comprehensive set of Terms of Service, including provisions on:

• Service access and permitted use
• Consultancy services
• Service levels and support
• Fees, payment, overage, and renewals
• Intellectual property
• Customer data and data protection
• Confidentiality
• Warranties and liability
• Indemnities
• Termination and governing law

This substantially expands the Customer’s contractual obligations and Dixa’s rights.

2. Customer Data and AI-Related Use

AI products and usage-based billing

“Add-On” now expressly includes AI products such as:

• Mim AI Co-Pilot
• AI Quality Assurance
• Other AI-related modules

The definition of “Minimum Commitment” includes usage-based subscriptions, including “conversations handled by AI agents.” “Overage” includes exceeding limits for analyzing data with AI tools or services. This creates potential additional charges based on AI usage, with overage charges generally non-cancellable and non-refundable.

Use of aggregated and anonymized data

Dixa receives an express, continuing right—during and after the Agreement—to:

• Aggregate, collect, and analyze data and information relating to the provision, use, and performance of the Services, including traffic and telemetry data;
• Use that information to develop and improve the Services and other Dixa offerings; and
• Disclose it in aggregated and anonymized form that does not identify the Customer or any individual.

Dixa also states that it will obtain and retain rights to this aggregated and anonymized data.

AI training risk: The clause does not expressly mention “training AI models,” machine learning, or generative AI. However, “develop and improve the Services and other Dixa offerings” is broad enough to potentially cover model development or training using qualifying aggregated/anonymized data. The clause does not clearly explain:

• What data is included;
• What anonymization standard applies;
• Whether Customer Data or message content may be used;
• Whether derived datasets, embeddings, prompts, outputs, or model parameters may be retained;
• Whether data is shared with AI vendors or subprocessors; or
• Whether the Customer can opt out.

The clause says data will not permit identification of an individual and will not identify the Customer, but it does not expressly exclude commercially sensitive content or confidential business information from the analytics process.

Confidentiality interaction

Customer Data is included within the definition of Confidential Information, but the aggregated/anonymized-data provision creates a specific exception permitting Dixa’s continued use. Customers should not assume that confidentiality alone prevents use of de-identified data for product development or possible AI-related purposes.

3. Data Protection and Sensitive Data Restrictions

The Customer is designated as data controller, with Dixa’s standard Data Processing Agreement incorporated by reference. Unless separately agreed, Customers may not use the Services for special-category personal data under GDPR. Services are also stated not to be HIPAA or PCI DSS compliant.

Dixa may use equipment and service providers outside the EU, subject to an approved transfer mechanism such as Standard Contractual Clauses.

4. Commercial and Operational Risks

• Automatic renewal: 12-month renewal terms, with at least three months’ notice required for non-renewal.
• Minimum commitments: Commitments generally cannot be reduced during the term, even if actual usage falls.
• Price increases: Automatic compounded 5% annual indexation, plus possible immediate changes to list and overage prices.
• Broad suspension rights: Dixa may suspend accounts for breach or non-payment.
• Unilateral amendments: Dixa may modify the Terms, generally with 30 days’ notice; continued use constitutes acceptance.
• Limited liability: Aggregate liability is capped at fees paid during the preceding six months, with broad exclusions for data loss and indirect damages.
• Data loss remedy: Dixa’s primary obligation is commercially reasonable restoration from its latest backup.

5. Recommended Review Points

Customers should review the incorporated Data Processing Agreement and AI product terms, seek an express prohibition or opt-out from using Customer Data for AI training, clarify anonymization and retention standards, and confirm whether AI-generated outputs and derived data belong to the Customer.

Summary

The supplied diff states only: “Added approximately 23 words to the document.” It does not include the actual wording of the added text or identify where it was inserted.

Legal and Commercial Impact

• Substantive changes cannot be assessed without the exact 23 words and their location in the agreement.
• It is not possible to determine whether the addition changes:
• Customer obligations or liability
• Provider rights to use customer data
• Confidentiality or security commitments
• Intellectual-property ownership
• Data retention or deletion
• Indemnities, warranties, or limitations of liability
• Termination rights or governing law

AI Model Training and Data Use

The provided diff contains no identifiable language concerning AI, machine learning, model training, automated processing, or use of customer data. Accordingly, no conclusion can be drawn about whether the amendment:

• Permits or restricts training AI models using customer data
• Allows use of customer prompts, inputs, outputs, or usage metadata for training
• Applies only to de-identified or aggregated data
• Requires customer consent or provides an opt-out
• Grants the provider ownership or a license to customer data
• Allows disclosure of data to affiliates or third-party AI providers

Recommended Next Step

Provide the actual added wording, preferably with the surrounding clause and any deleted or replacement text. The exact language is necessary to identify new legal risks, particularly any expanded rights to use customer data for AI training.

Summary of Important Changes

1. New Terms of Service and major restructuring

• The Terms are now dated 28 May 2025 and appear to replace the prior version in full.
• The agreement now expressly includes more detailed definitions, service rules, billing terms, data provisions, intellectual-property terms, and termination provisions.
• The agreement applies when the customer accesses or uses the Services, authorizes an agent or end user to do so, or accepts an Order Form.

2. AI products and AI-related billing

New definitions expressly cover AI-related products, including:

• “Mimic AI Co-Pilot”
• “AI Quality Assurance”
• Other Add-Ons and AI-agent functionality.

The agreement also introduces AI-related usage charges:

• “Overage” includes analyzing more data with AI tools than included in the applicable plan.
• Minimum Commitments may include usage-based subscriptions, including the number of conversations handled by AI agents.
• AI Add-Ons may be priced by Agent seats and may generate additional overage charges.
• Overage is billed at then-current rates, is non-cancellable and non-refundable, and may be changed through the platform.

Risk: Customers should confirm what AI features are enabled, what data they process, included usage limits, and the applicable overage rates. The contract may permit additional charges based on AI usage without a discount or refund.

3. Use of Customer Data to improve services and AI

The revised confidentiality clause expressly permits Dixa to:

• Aggregate, collect, and analyze data and information relating to the provision, use, and performance of the Services, including traffic and telemetry data.
• Use that information during and after the agreement to develop and improve the Services and other Dixa offerings.
• Disclose it only in aggregated and anonymized form that does not identify the Customer or any individual.
• Retain rights in the aggregated and anonymized data.

There is no express statement that identifiable Customer Data may be used to train AI models. However, the wording is broad enough to permit use of aggregated and anonymized data for product improvement, which could include improving or developing AI systems unless separately restricted.

Risks and points to clarify:

• “Aggregated” and “anonymized” are not defined, and the contract does not specify the anonymization standard or prohibit re-identification.
• The clause applies during and after termination.
• It does not expressly exclude conversation content, prompts, outputs, recordings, or other Customer Data from model-development use if Dixa considers them sufficiently anonymized.
• Customers requiring a strict no-training commitment should seek an express written restriction covering Customer Data, prompts, inputs, outputs, recordings, and metadata.

4. Customer responsibility for data

• The Customer retains ownership of Customer Data but bears responsibility for its legality, accuracy, integrity, and third-party rights.
• Customers generally may not process sensitive personal data unless expressly agreed in the data processing agreement.
• Services may be performed outside the EU, subject to an approved transfer mechanism.

5. Commercial and termination risks

• Annual Fees automatically increase by 5%, compounded annually, without separate notice.
• Subscriptions generally cannot be downgraded or cancelled during the current term and automatically renew for successive 12-month periods unless notice is given at least three months before expiry.
• Dixa may change the Services and amend the Terms with notice; continued use constitutes acceptance.
• Dixa’s aggregate liability is capped at Fees paid during the preceding six months, with broad exclusions for data loss and consequential losses.

Summary of Important Changes

1. Major restructuring and expanded definitions

The revised Terms add a comprehensive definitions section and incorporate or link to numerous legal documents, including the Privacy Policy, DPA, security information, telecommunication terms, and Early Access Terms (“EAT”).

Important new definitions include:

• Add-Ons, including “Mim AI Co-Pilot,” “AI Quality Assurance,” Collaboration Users, and Advanced Insights.
• Customer Data, defined broadly to include data supplied by the Customer, uploaded by users, and content provided by Customer End-Users.
• Minimum Commitment, which expressly includes AI-agent usage and other feature- or usage-based subscriptions.
• Overage, which includes exceeding limits for AI analysis, API calls, Agents, communications, or other usage.

Risk: The Customer may be contractually committed to minimum AI or other usage levels and may incur uncapped or variable overage charges.

2. AI and use of Customer data

The revised Terms do not expressly state that Dixa may use identifiable Customer Data or customer conversations to train general-purpose AI models. However, they introduce AI products and broader data-use language.

Dixa may:

• Aggregate, collect, and analyze data relating to the provision, use, and performance of the Services, including traffic and telemetry data.
• Use that data during and after the Agreement to develop and improve the Services and other Dixa offerings.
• Disclose it only in aggregated and anonymized form that does not identify the Customer or an individual.

The clause is limited to data handled “in a way that does not permit identification,” but it does not expressly address:

• Whether Customer Content or conversations may be used for machine-learning or model training;
• Whether prompts, outputs, transcripts, or feedback are retained;
• Whether data is used to train shared or third-party models;
• Re-identification safeguards or technical anonymization standards; or
• Customer opt-out rights.

Risk: Although the wording does not clearly authorize training on identifiable Customer Data, “develop and improve” is broad enough to create uncertainty about whether de-identified data, usage patterns, or AI interactions may be used for model development. Customers requiring a strict prohibition on AI training should seek an express contractual restriction.

3. Data protection and international processing

The Customer is expressly treated as the data controller, with Dixa’s standard DPA incorporated unless a separate DPA applies. Services may be performed outside the EU, subject to an adequacy mechanism such as Standard Contractual Clauses.

The Terms also state that Customers should not process special-category personal data unless specifically agreed in the DPA, and that the Services are not HIPAA- or PCI-DSS-compliant.

4. Commercial and billing changes

The revised Terms introduce or clarify:

• A 5% annual compound fee indexation, applied automatically without separate notice.
• Minimum commitments that cannot generally be reduced during the Subscription Term.
• Overage billing for AI analysis, Agents, communications, and other usage.
• Non-refundable and non-cancellable overage charges.
• Automatic monthly card billing and immediate-effect changes to published overage prices.
• Twelve-month automatic renewals with three months’ advance notice required for non-renewal.

Risk: Costs may increase through annual compounding, minimum commitments, and usage-based AI charges, with limited ability to downgrade or cancel mid-term.

5. Liability and data-loss position

Dixa’s aggregate liability is capped at fees paid during the preceding six months. For Customer Data loss, the exclusive remedy is commercially reasonable restoration from the latest backup. Indirect losses, including data corruption and lost profits, are excluded.

6. Other notable changes

The Customer must:

• Ensure individual, non-shared User Accounts;
• Comply with changing Documentation and operating policies;
• Accept Dixa’s right to suspend accounts for violations;
• Participate in marketing activities and permit use of its name and logo;
• Accept Dixa’s ability to modify the Services and amend the Terms with notice.

Summary of Important Changes

AI and Customer Data Use

• New express permission to use aggregated and anonymized data: Dixa may “aggregate, collect and analyse” data relating to the provision, use and performance of the Services, including traffic and telemetry data.
• Permitted purposes: Dixa may use this data during and after the Agreement to:
• develop and improve the Services; and
• develop and improve other Dixa offerings.
• Disclosure limitation: Dixa may disclose the information only in aggregated and anonymized form that does not identify the Customer or any individual.
• Ownership/rights: Dixa will obtain and retain rights to the aggregated and anonymized data.
• AI-specific additions: The definitions now expressly refer to AI products, including “Mim AI Co-Pilot” and “AI Quality Assurance.” “Overage” also expressly includes analyzing more data with AI tools than included in the applicable plan.

AI-related risk

The clause does not expressly authorize use of identifiable Customer Data, customer conversations, or personal data to train general-purpose AI models. However, it permits broad use of aggregated and anonymized data to improve “other Dixa offerings,” which could include AI products or models. The Agreement does not define the anonymization standard, prohibit model training, explain whether data will be irreversibly anonymized, or give the Customer an opt-out. Customers handling confidential, regulated, or sensitive data should seek clarification or a written restriction covering AI training and model development.

Commercial and Operational Changes

• Minimum commitments: Minimum Commitments now expressly include Agent seats, add-ons, telecommunication packages, and usage-based services such as AI-agent conversations. Commitments generally cannot be reduced during the Subscription Term.
• Automatic renewal: Agreements renew for successive 12-month periods unless notice is given at least three months before expiry.
• Price increases: A 5% annual, compounded indexation applies automatically to subscription fees without separate notice. Dixa may also immediately update list and Overage prices through the Platform.
• Overage charges: Excess usage is billed at then-current rates, without discount, and is non-cancellable and non-refundable.
• Service changes: Dixa may change the Services and Documentation as it considers appropriate, potentially affecting functionality.
• Unilateral amendments: Dixa may amend the Terms, generally with 30 days’ notice, and continued use constitutes acceptance.

Liability and Data Risks

• Dixa’s aggregate liability is capped at fees paid in the six months preceding the relevant incident.
• The Customer’s remedy for lost or damaged data is limited to commercially reasonable restoration efforts from Dixa’s latest backup.
• Dixa disclaims responsibility for data loss caused by third parties and for third-party integrations.
• Customers are responsible for ensuring the legality, accuracy, and third-party rights relating to Customer Data.

Summary of Important Changes

1. Major restructuring and expansion of the Terms

The revised document substantially replaces the previous introductory and definitions sections and introduces a much longer, more detailed agreement. It now expressly incorporates or links to additional legal documents, including the Privacy Policy, GDPR materials, Data Processing Agreement, Data Security information, Service Level Agreement, Early Access Terms, and Referral Program terms.

The agreement is also more clearly defined as a binding contract, including where services are accessed without a signed Order Form.

2. New and expanded product definitions

The revised Terms introduce detailed definitions for:

• Add-Ons, including Mim AI Co-Pilot, AI Quality Assurance, Collaboration User, and Advanced Insights.
• AI-related usage, including “conversations handled by AI agents” and data analyzed by AI tools.
• Minimum Commitment, which may include agent seats, add-ons, telecommunications packages, and AI-agent usage.
• Overage, covering excess AI analysis, API calls, communications usage, agents, and other usage limits.
• Admin Interface, allowing administrators to purchase additional services, seats, numbers, and add-ons.

These provisions create greater potential for usage-based charges and make minimum commitments difficult to reduce during the subscription term.

3. AI models and use of customer data

The most important AI-related change appears in the confidentiality section. Dixa may:

• Aggregate, collect, and analyze data and information relating to the provision, use, and performance of the Services, including traffic and telemetry data.
• Use that information during and after the agreement to develop and improve the Services and other Dixa offerings.
• Disclose the information in aggregated and anonymized form, provided it does not identify the Customer or an individual.

The clause does not expressly authorize Dixa to train AI models on identifiable Customer Data or customer content. However, “data and other information relating to” service provision and performance is broad and could include product usage data, telemetry, and potentially information derived from AI features. The provision is not limited to model training, nor does it specify retention periods, opt-out rights, or clear technical standards for anonymization. Customers should seek clarification on whether prompts, conversations, transcripts, outputs, or other Customer Data are used to train or improve AI models.

The revised Terms also define AI products and permit AI-related overage billing, but do not provide detailed AI-specific protections, accuracy commitments, human-review requirements, or restrictions on provider-side model training.

4. Commercial and billing risks

New or expanded provisions include:

• Automatic 5% annual compounded fee indexation, without separate notice.
• Minimum commitments that cannot generally be reduced during the subscription term.
• Immediate or monthly billing for overage at then-current rates, with no discount and no refund once incurred.
• AI analysis and AI-agent activity expressly included as potentially billable overage.
• Pricing and overage rates potentially changing through the Platform.
• Automatic renewal for successive 12-month terms, requiring at least three months’ notice of non-renewal.
• Upgrades taking effect immediately, while downgrades generally apply only at renewal.

5. Data protection and confidentiality

The Customer remains responsible as data controller for Customer Data and warrants its legality and accuracy. Use of sensitive personal data is prohibited unless specifically agreed in the DPA. Services may be performed outside the EU, subject to an approved transfer mechanism.

Dixa’s liability for data loss is limited primarily to commercially reasonable restoration from its latest backup. Confidentiality obligations apply, but the new aggregated/anonymized data-use right survives termination.

6. Liability and operational risks

Dixa disclaims warranties that the Services will be uninterrupted, error-free, or fit for purpose. Aggregate liability is capped at Fees paid during the preceding six months, while indirect losses—including data loss—are excluded. Dixa may modify the Services and Terms, generally with 30 days’ notice for material changes, and may suspend access for breaches or non-payment.

Structured Summary of Important Changes

1. New or expanded Terms of Service framework

• The document is now expressly dated 28 May 2025 and states that the Terms have been updated.
• The agreement is more fully defined, including the Order Form, online signup, referenced documents, Documentation, Privacy Policy, and Data Processing Agreement.
• The Terms introduce or expand definitions for:
• Add-Ons, including Mim AI Co-Pilot and AI Quality Assurance
• AI-related usage, including data analysed by AI tools and conversations handled by AI agents
• Collaboration Users, Customer Data, Overages, Minimum Commitments, and Number Subscriptions

2. AI and customer-data use

The revised Terms contain an express data-use permission:

• Dixa may aggregate, collect, and analyse data and information relating to the provision, use, and performance of the Services, including traffic and telemetry data.
• Dixa may use that information during and after the agreement to:

1. Develop and improve the Services and other Dixa offerings; and

2. Disclose it only in an aggregated and anonymised format that does not identify the Customer or any individual.

• Dixa states that it will obtain and retain rights in the aggregated and anonymised data.

Risk and interpretation

• This is the principal AI/data-use change. It appears to permit use of service-related data for product development and improvement, which could include improving AI functionality, although the Terms do not expressly say that Customer Data may be used to train generative AI models.
• The permission is limited by the wording that the data must not permit identification of an individual for the stated aggregation/anonymisation and disclosure purposes. However, the clause does not define the anonymisation standard or expressly exclude Customer Data, prompts, conversations, transcripts, or outputs from model-training activities.
• Customers requiring a strict prohibition on AI training should seek clarification or a contractual opt-out in the Order Form or Data Processing Agreement.
• The confidentiality clause otherwise limits use of Confidential Information to performing the agreement, but the new aggregated/anonymised-data clause operates as an express exception.

3. AI pricing and usage exposure

• Minimum Commitments may include usage-based subscriptions, including conversations handled by AI agents.
• Overages may arise from:
• Analysing more data with AI tools than included in a plan; and
• Exceeding other AI-related quantitative limits.
• Overages are charged at then-current rates, are non-cancellable and generally non-refundable, and may be billed monthly.
• Minimum Commitments cannot generally be reduced during the Subscription Term, even if actual AI usage is lower.

4. Commercial and operational risks

• Automatic 5% annual compounded fee indexation applies to subscriptions.
• The agreement automatically renews for successive 12-month terms unless notice is given at least three months before expiry.
• Dixa may change Services and Documentation and may amend the Terms with 30 days’ notice, subject to stated exceptions.
• Dixa may suspend access for breach or non-payment.
• Liability is broadly limited to the Fees paid in the preceding six months, with broad exclusions for indirect loss and data loss.
• The Customer bears responsibility for Customer Data legality, third-party rights, and use by its users and affiliates.

Summary of Important Changes

1. New and Expanded Definitions

The Terms now include a substantially expanded definitions section, including:

• Add-Ons, expressly including AI products such as Mim AI Co-Pilot and AI Quality Assurance.
• Customer Data, defined broadly to include data supplied by the Customer, uploaded by users, and content provided by Customer End-Users.
• Overage, including usage of AI tools, AI data analysis, API calls, messages, voice minutes, and excess users.
• Minimum Commitment, which may include AI-agent conversations, Add-On subscriptions, seats, telecommunications packages, and usage-based services.

Risk: The broad definitions make AI usage potentially chargeable and contractually binding even where actual usage is lower than expected. Minimum commitments generally cannot be reduced during the subscription term.

2. AI and Customer Data Use

The most relevant data-use provision states that Dixa may:

• Aggregate, collect, and analyze data and information relating to the provision, use, and performance of the Services, including traffic and telemetry data.
• Use that information during and after the Agreement to develop and improve the Services and other Dixa offerings.
• Disclose the information only in an aggregated and anonymized format that does not identify the Customer or any individual.

Dixa also claims rights in the resulting aggregated and anonymized data.

AI-specific implications

• The Terms expressly contemplate AI products and AI-related usage charges.
• The Terms do not expressly authorize use of identifiable Customer Data, customer conversations, prompts, outputs, or Content to train general-purpose AI models.
• However, “data and other information relating to” the Services is broad and could potentially include derived analytics, usage information, or other AI-related performance data.
• The anonymization standard is described as information that does not identify the Customer or any individual, but the Terms do not specify the anonymization methodology, re-identification safeguards, retention period, or whether human review is permitted.
• The data-use clause applies during and after termination, with no stated deletion deadline for aggregated or anonymized data.

Recommended clarification: Require an express statement that Customer Data, Content, conversations, prompts, outputs, and personal data will not be used to train or fine-tune general-purpose or third-party AI models, except with the Customer’s documented opt-in consent. The contract should also define anonymization, prohibit re-identification, and identify permitted subprocessors and retention periods.

3. Data Protection and International Transfers

The Customer is designated as data controller, and Dixa’s standard Data Processing Agreement applies unless separately replaced. The Terms:

• Restrict processing of sensitive personal data unless specifically agreed in the DPA.
• Permit services and equipment outside the EU.
• Require non-EU/EEA providers to use an approved transfer mechanism, such as Standard Contractual Clauses.

Risk: Customers must review the separate DPA and trust-centre materials because important AI processing, subprocessors, security, retention, and international-transfer details may be outside the Terms.

4. Commercial and Operational Risks

• Automatic 5% annual compounded fee indexation applies to subscriptions.
• Pricing and Overage rates may change; list and Overage prices may change immediately through the Platform.
• Subscriptions generally cannot be downgraded or cancelled during the term.
• Contracts automatically renew for successive 12-month periods unless notice is given at least three months before expiry.
• Dixa may suspend accounts for breaches or non-payment.
• Dixa may modify the Services and Documentation over time.

Structured Summary of Important Changes

1. New Terms and Expanded AI-Related Products

• The Terms are reissued as “Terms of Service – 28 May 2025”, replacing the prior presentation and terminology.
• New definitions and commercial provisions expressly address:
• AI Add-Ons, including Mim AI Co-Pilot, AI Quality Assurance and Advanced Insights.
• AI-agent usage, including “conversations handled by AI agents” as a possible Minimum Commitment.
• AI tools and services as usage categories subject to Overage charges.
• Risk: Customers may be committed to minimum volumes and charged for usage exceeding limits, including AI analysis and AI-agent activity. The terms state that minimum commitments generally cannot be reduced during the Subscription Term.

2. Use of Customer Data for AI and Service Improvement

• The confidentiality section now expressly permits Dixa to aggregate, collect and analyse data and other information relating to the provision, use and performance of the Services, including traffic and telemetry information.
• Dixa may use this information, during and after the Agreement, to:

1. Develop and improve the Services and other Dixa offerings; and

2. Disclose it only in an aggregated and anonymised format that does not identify the Customer or any individual.

• Dixa is entitled to retain rights in the aggregated and anonymised data.

Important AI-training implication:

The wording does not expressly say that Dixa may train AI models on identifiable Customer Data, communications, transcripts or prompts. However, “develop and improve” the Services and “other Dixa offerings” is broad enough to cover model development or training using aggregated/anonymised datasets. The Terms do not define the anonymisation standard, prohibit re-identification, or specify whether Customer Data may be used to train third-party or general-purpose models.

Risk: Customers seeking to prohibit use of their data for AI training may need a negotiated DPA, Order Form, or written opt-out. The provision also applies after termination.

3. Customer Data Ownership and Privacy Responsibilities

• Customer retains ownership of Customer Data and remains responsible for its legality, accuracy and integrity.
• The Customer is designated as data controller, with Dixa’s standard DPA incorporated unless a separate DPA applies.
• Use of sensitive personal data is prohibited unless specifically agreed in the DPA.
• Services may be operated outside the EU, using approved transfer mechanisms such as Standard Contractual Clauses.

Risk: The main Terms impose broad customer responsibility while the detailed limits on Dixa’s processing of AI-related data appear to be left to the DPA and other documents.

4. Commercial and Operational Changes

• New and expanded definitions cover collaboration users, telephony, SMS/MMS, overages, subscriptions and minimum commitments.
• Dixa may update list and Overage prices immediately by posting them in the Platform.
• A compounded 5% annual fee indexation applies automatically.
• Subscriptions generally cannot be cancelled or downgraded during the current term.
• Dixa may suspend accounts for breach or non-payment and may modify Services and Documentation.

5. Liability and Data Loss

• Dixa’s aggregate liability is capped at fees paid during the preceding six months.
• For lost or damaged Customer Data, the exclusive remedy is commercially reasonable restoration from Dixa’s latest backup.
• Dixa disclaims responsibility for loss caused by third parties and excludes liability for indirect losses, including data loss.

Overall assessment: The revised Terms create a clearer framework for AI products and usage billing, but the AI-data language is broad and leaves important safeguards—such as model-training restrictions, deletion, retention, opt-out rights and re-identification standards—to the DPA or separate negotiation.

Contract Diff Analysis

Key Limitation

The supplied diff states only:

> “Added approximately 9125 words to the document”

It does not include the actual added language, deleted language, or replacement provisions. As a result, it is not possible to reliably identify:

• New legal obligations or rights
• Changes to liability, indemnification, warranties, or termination
• Changes to confidentiality or security requirements
• Changes to governing law or dispute resolution
• Changes concerning the customer’s data
• Whether customer data may be used to train, fine-tune, evaluate, or improve AI models

AI Training and Data-Use Review

No conclusion can be drawn from the available information about whether the new text:

• Permits the provider to use customer data to train AI models
• Limits training use to anonymized, aggregated, or de-identified data
• Allows use of prompts, inputs, outputs, metadata, logs, or support data
• Grants the provider a license or ownership interest in customer data or outputs
• Allows data to be shared with affiliates, subcontractors, or third-party model providers
• Retains data after termination for model training or other purposes
• Provides an opt-out, deletion right, or customer-controlled setting
• Uses customer data for human review, quality assurance, safety testing, or product improvement
• Restricts use of personal, confidential, regulated, or sensitive information

Risk Assessment

Because the substantive language is missing, the risks are undetermined. The statement that approximately 9,125 words were added is itself significant: a large addition may contain material changes that are not merely editorial. It should not be assumed that the additions are favorable or administrative.

Information Needed

Please provide the complete redlined text, including:

1. All additions, deletions, and replacements

2. The relevant prior version and revised version, if available

3. Any sections addressing data use, privacy, AI, confidentiality, security, intellectual property, or service improvement

Once provided, the changes can be analyzed for commercial impact, legal risk, and—specifically—whether customer data may be used to train or improve AI models.

Key Changes and Risks

1. New Terms and Structure

• The Terms are dated 28 May 2025 and appear to replace the prior Terms of Service in full.
• The revised Terms introduce a substantially expanded definitions section and expressly incorporate:
• Order Forms;
• Documentation;
• Dixa’s Data Processing Agreement (DPA);
• Privacy Policy and other policies; and
• Additional telecommunications terms where applicable.
• The Agreement is formed through access or use of the Services, including by agents or end-users, not only by signature.

2. AI Features and AI-Related Charges

• “Add-On” now expressly includes AI products such as Mim AI Co-Pilot and AI Quality Assurance.
• “Minimum Commitment” expressly includes usage-based services, including conversations handled by AI agents.
• “Overage” includes:
• analysing more data with AI tools than included in the purchased plan;
• exceeding usage limits for AI agents; and
• additional Agents using paid AI Add-Ons.
• AI-related overage is billable at then-current rates, generally without discounts, and is non-cancellable and non-refundable once incurred.
• The Customer may be committed to minimum AI-related volumes even if actual usage is lower.

Risk: Customers should confirm how AI usage is measured, what counts as an analysed conversation or data unit, applicable limits, and whether AI overage pricing can change immediately through the Platform.

3. Use of Data to Improve Services and AI Models

A new confidentiality exception permits Dixa to:

• aggregate, collect and analyse data and information relating to the provision, use and performance of the Services, including traffic and telemetry;
• use that information during and after the Agreement to develop and improve the Services and other Dixa offerings; and
• disclose it only in aggregated and anonymised form that does not identify the Customer or an individual.

The clause does not expressly state that Customer Data, prompts, conversations, transcripts, or outputs may be used to train AI models. It also does not expressly prohibit such use where Dixa considers the data aggregated or anonymised.

Risk: The language may permit model-development or AI-training uses indirectly, particularly for de-identified or aggregated conversation data. Customers requiring a strict prohibition on training should seek an express contractual restriction covering Customer Data, prompts, transcripts, metadata, and outputs.

4. Data Protection and Confidentiality

• The Customer is designated as data controller, with Dixa’s standard DPA applying unless replaced by a separate DPA.
• Sensitive personal data is prohibited unless specifically agreed in the DPA.
• Services may be operated outside the EU/EEA using approved transfer mechanisms.
• Confidential Information must generally be returned or destroyed on termination, but the new aggregated/anonymised data right survives termination.

Risk: The DPA and Privacy Policy should be reviewed for AI-specific processing, subprocessors, retention, deletion, and international transfer terms.

5. Commercial and Operational Changes

• Automatic 5% annual compounded fee indexation applies to subscriptions.
• Minimum commitments cannot generally be reduced during the term.
• Subscriptions automatically renew for successive 12-month periods unless notice is given at least three months before expiry.
• Dixa may change Services and Documentation and may modify Terms with notice.
• Liability is capped at Fees paid in the preceding six months, with broad exclusions for data loss and consequential loss.

Summary

The provided diff does not include the actual contractual language. It only states:

> “Added approximately 9,125 words to the document”

Accordingly, it is not possible to reliably identify:

• Changes to customer data rights or permitted uses
• Whether customer data may be used to train, fine-tune, or evaluate AI models
• Whether data may be used to improve services or develop products
• Data retention, deletion, anonymization, or aggregation provisions
• Disclosure or sharing of data with affiliates, vendors, or AI providers
• Ownership of inputs, outputs, feedback, or model-generated materials
• Confidentiality, security, or breach obligations
• Indemnities, liability limitations, audit rights, or termination effects
• Any newly introduced customer obligations or reduced provider obligations

Potential Risk

The addition of approximately 9,125 words may represent substantial new obligations or expanded provider rights. However, the nature and significance of those changes cannot be assessed without the inserted text and any surrounding language showing what was removed or replaced.

AI-Training Review

No conclusion can be reached regarding AI-model training. The diff does not reveal whether the agreement now permits the provider to:

• Train general-purpose or customer-specific AI models using customer data
• Use prompts, inputs, outputs, or metadata for model improvement
• Retain data for training after termination
• Share data with third-party model providers
• Use de-identified, aggregated, or derived data for AI development
• Opt the customer in or out of such uses

Information Needed

Please provide the full redline or the actual 9,125 words added, including the relevant deleted and replacement text. The AI-data provisions should be reviewed together with definitions of “Customer Data,” “Usage Data,” “Content,” “Inputs,” “Outputs,” “Aggregated Data,” and “Services.”

Structured Summary of Important Changes

1. Document title, navigation, and introductory wording

• The document heading has been expanded to include links or references to various Legal Hub materials, including:
• Terms of Service
• Cookie Policy
• GDPR
• Data Processing Agreement
• Data Security
• Applicant Privacy Policy
• Accessibility Statement
• Early Access Terms
• Referral Program
• The revised heading also adds “Last updated: 25 Novembre 2021.”
• The opening sentence now expressly states that the Privacy Policy is incorporated into and subject to the Dixa Terms of Service, rather than referring only to the “Dixa Terms.”
• “Agreement” is now defined using standard quotation marks, and the revised text clarifies that capitalized terms have the meanings assigned to them in the Policy.

Risk/impact: The expanded references may make additional linked documents relevant to the customer’s contractual relationship. Customers should check whether those documents contain separate data-use, security, or liability terms. The changes appear primarily editorial, but the incorporation by reference of the Terms of Service should be noted.

2. Defined terms and grammatical revisions

• The definitions of “Dixa,” “us,” “we,” “Website,” “Dixa Services,” and “Customer Data” have been reformatted with corrected quotation marks and punctuation.
• “Personal Data” is now consistently presented as a defined term.
• Numerous apostrophe, punctuation, and wording corrections were made, including:
• “Dixa’s” → “Dixa's”
• “third-party’s” → “third-party's”
• “customers’” → “customers'”
• “opt-out” and “Do Not Track” formatting
• Minor grammar and spacing corrections

Risk/impact: No apparent substantive change to the scope of Customer Data or Personal Data is shown in the diff.

3. Data-use and disclosure provisions

• The revised text retains existing statements that information may be used for:
• Analyzing usage trends and preferences
• Administering and optimizing the Dixa Services
• Monitoring usage or traffic patterns
• Understanding customer and visitor interests, habits, and usage patterns
• Complying with legal, bookkeeping, shareholder, and corporate obligations
• The text continues to state that Personal Data forming part of Customer Data is subject to the Customer’s own privacy policy and that affected individuals generally must direct access, correction, or deletion requests to the Customer.

4. AI model training

• No express provision concerning artificial intelligence, machine learning, model training, model improvement, or use of Customer Data to train AI models appears in the provided diff.
• The general language concerning analyzing usage, optimizing services, and understanding usage patterns remains broad, but the diff does not show it being expanded to expressly authorize AI training.
• The absence of an AI-training provision does not necessarily establish that AI-related processing is prohibited; other incorporated documents, the Terms of Service, DPA, or product-specific terms may address it.

5. Hosting, security, and changes

• Existing language remains regarding EU customers’ databases being hosted in an EU data center, while Dixa Services may otherwise be hosted in the United States or EU.
• Existing breach-notification and security language appears unchanged apart from formatting.
• The procedure for posting policy changes and indicating a new effective date remains substantially unchanged.

Overall assessment

The diff appears predominantly editorial and structural. No clear new authorization to use Customer Data for AI-model training is introduced in the supplied text. The main contractual point is the clearer incorporation of the Dixa Terms of Service and the expanded references to related legal documents.

Summary

The diff indicates that approximately 23 words were added, but it does not include the actual wording of those additions.

Legal and Commercial Impact

• The specific legal effect cannot be determined without seeing the added language.
• It is not possible to assess whether the additions:
• Change the parties’ rights or obligations;
• Create new fees, liability, indemnity, confidentiality, or termination risks;
• Modify governing law, dispute resolution, or compliance requirements; or
• Expand permissions to use, share, retain, or commercialize customer data.

AI Training and Customer Data

• The provided diff does not reveal whether customer data may be used to train, fine-tune, evaluate, or improve AI models.
• It is therefore not possible to determine:
• Whether customer data or prompts may be used for model training;
• Whether training is automatic or requires customer consent;
• Whether data is anonymized, aggregated, or de-identified;
• Whether human reviewers or third-party providers may access the data;
• Whether customer data may be used to train models benefiting other customers; or
• How long training data is retained or how deletion requests are handled.

Conclusion

The actual 23-word addition is required for a meaningful legal analysis. Based solely on the description, no specific new risk can be identified.

Structured Summary of Important Changes

1. New/reissued Terms of Service

• The document is substantially recast as “Terms of Service — 28 May 2025” rather than the prior “Terms of Service”/“EAT” presentation.
• The revised terms apply when a customer accesses or uses the Services, authorizes an agent or end user to do so, or accepts an Order Form.
• The agreement now expressly includes the Order Form and referenced documents, including the Documentation and Dixa’s Data Processing Agreement (“DPA”).

2. New product, user and billing definitions

The revised terms add or clarify definitions for:

• AI and other Add-Ons, expressly including “Mim AI Co-Pilot” and “AI Quality Assurance.”
• AI usage, including “analysing more data with AI tools or services” as a potential overage.
• Minimum Commitment, which may include Agent seats, Add-Ons, telecommunications packages and usage-based services such as conversations handled by AI agents.
• Collaboration Users, replacing “View-Only Users.”
• Overage, which is charged at then-current rates and is generally non-cancellable and non-refundable.

Customer risk

Minimum commitments cannot generally be reduced during the term, even if actual usage is lower. AI and other usage-based features may therefore create additional charges, including charges based on usage above the contracted threshold.

3. Customer Data and data protection

• The customer retains ownership of Customer Data and remains responsible for its legality, accuracy, integrity and quality.
• The customer is designated as the data controller, with Dixa acting under its standard DPA unless a separately negotiated DPA applies.
• Use of special-category/sensitive personal data is prohibited unless specifically agreed in the DPA.
• Services and processing may occur outside the EU, subject to an approved transfer mechanism such as Standard Contractual Clauses.

4. AI model training and use of data

The revised confidentiality provisions expressly allow Dixa to:

• Aggregate, collect and analyse data and information relating to the provision, use and performance of the Services, including traffic and telemetry data;
• Use that information during and after the contract to develop and improve the Services and other Dixa offerings; and
• Disclose it only in aggregated and anonymised form that does not identify the customer or an individual.

AI-specific risk

The provision does not expressly mention “training AI models,” but “develop and improve” the Services and other offerings could reasonably encompass AI model development or training. The clause does not provide:

• A customer opt-out;
• A precise definition or verification standard for “aggregated and anonymised” data;
• A clear prohibition on using customer-derived prompts, conversations, transcripts or outputs before anonymisation; or
• A specific deletion period after termination.

Dixa also states that it obtains and retains rights in the aggregated and anonymised data.

5. Commercial and operational changes

• Fees are subject to an automatic 5% compounded annual indexation, in addition to any prior increases.
• Dixa may change list and overage prices immediately by posting them in the Platform.
• Subscriptions automatically renew for successive 12-month terms unless notice is given at least three months before expiry.
• Dixa may modify the Services and Terms, generally with 30 days’ notice, although continued use constitutes acceptance.
• Dixa may suspend access for breach or non-payment and disclaims responsibility for third-party integrations.

Summary of Important Changes

1. Major restructuring and expanded definitions

The Terms have been substantially reorganized and expanded. New definitions include:

• Add-Ons, including Mim AI Co-Pilot and AI Quality Assurance.
• Customer Data, expressly covering data uploaded by users and content provided by customer end-users.
• Overage, including excess use of AI tools, API calls, communications, and agents.
• Minimum Commitment, including agent seats, add-ons, telecommunications packages, and AI-agent conversation usage.
• Admin Interface, Collaboration Users, Dixa Materials, and related service concepts.

Risk: The expanded definitions may broaden what counts as Customer Data and increase the types of usage that can generate binding commitments or additional charges.

2. AI-related use of data

The Terms do not contain a clear, express statement that identifiable Customer Data will be used to train general-purpose AI models. However, they introduce AI products and state that Dixa may:

• Monitor usage and charge for AI analysis beyond purchased limits.
• Aggregate, collect, and analyze data relating to the provision, use, and performance of the Services, including traffic and telemetry data.
• Use that data during and after the Agreement to “develop and improve the Services and other Dixa offerings.”
• Disclose the data only in aggregated and anonymized form that does not identify the Customer or any individual.

AI/data risks

• “Develop and improve” is broad and could potentially include improving AI systems or models, although model training is not expressly mentioned.
• The restriction applies to data that does not permit identification of an individual, but the Terms do not explain the anonymization or de-identification standard, whether re-identification risks are tested, or whether Customer Data itself may be used before anonymization.
• There is no express customer opt-out, approval right, limitation to service-specific models, or prohibition on using Customer Data to train models serving other customers.
• The separate Data Processing Agreement, Privacy Policy, and documentation may contain additional AI/data-use terms and should be reviewed together with these Terms.

3. Financial and commitment changes

The Terms add or clarify significant commercial obligations:

• Minimum Commitments cannot generally be reduced during the Subscription Term, regardless of actual usage.
• AI usage and other excess usage may be charged as Overage at then-current rates.
• A compounding 5% annual fee indexation applies automatically, without separate notice.
• Dixa may immediately change list prices and Overage prices by posting them in the Platform.
• Subscriptions generally cannot be downgraded or cancelled during the current term.
• All subscriptions automatically renew for successive 12-month periods unless notice is given at least three months before expiry.

Risk: Customers may face increased charges, locked-in capacity, AI usage overages, and missed renewal deadlines.

4. Data protection and liability

The Customer remains responsible for the legality, accuracy, and ownership of Customer Data and generally acts as controller. Special-category personal data is prohibited unless specifically agreed in the DPA.

Dixa’s liability is capped at Fees paid during the preceding six months, while indirect losses—including loss or corruption of data—are excluded.

Risk: The liability cap may be low relative to potential confidentiality, data-loss, or AI-related claims.

5. Other notable changes

• Dixa may modify the Services and Documentation over time.
• Dixa may use subcontractors while remaining responsible for overall performance.
• Customers agree to participate in marketing and permit use of their name and logo.
• Customers must use supported, up-to-date browsers and applications.
• Data may be processed outside the EU/EEA using an approved transfer mechanism.

Summary of Changes

Scope of the Diff

• The diff states that approximately 9,125 words were removed from the document.
• No replacement language or surviving text is provided.
• Because the deleted provisions are not identified, it is not possible to determine which contractual rights, obligations, limitations, or protections were removed.

Customer Data and AI-Training Terms

• The provided diff contains no specific language addressing customer data, artificial intelligence, machine learning, model training, or use of customer content.
• Accordingly, it is not possible to determine whether the revised document:
• Permits or prohibits training AI models using customer data;
• Allows use of customer data to improve products or services;
• Requires customer consent for AI training;
• Limits use to aggregated, anonymized, or de-identified data;
• Provides opt-out rights;
• Restricts human review or automated processing;
• Addresses ownership of model inputs, outputs, or trained models; or
• Imposes data-retention, deletion, confidentiality, or security obligations relating to AI systems.

Potential Legal Risks

The removal of 9,125 words could materially affect the customer’s position, particularly if the deleted text included:

• Confidentiality and data-use restrictions;
• Ownership or licensing provisions;
• Privacy and data-protection obligations;
• Security commitments;
• Indemnities and liability limitations;
• Audit, deletion, or data-return rights;
• Service-level commitments;
• Restrictions on subcontractors or third-party providers; or
• Provisions governing AI or machine-learning use.

If AI-training restrictions were among the deleted provisions, the revised agreement could potentially give the provider broader rights to use customer data, including for product development or model training. However, that conclusion cannot be confirmed from the information supplied.

Recommended Next Step

Obtain either:

1. The full original and revised agreements; or

2. A redline identifying the specific deleted clauses.

The deletion-only description is insufficient to conduct a reliable legal comparison or assess the effect on customer-data and AI-training rights.