Datadog

Contract Diff Analysis

Summary

The supplied diff does not include the text of the new provisions. It only states:

> “Added approximately 4172 words to the document”

Because the actual additions, deletions, and replacements are not provided, it is not possible to reliably identify:

• New legal obligations or rights
• Changes to liability, indemnity, confidentiality, or termination provisions
• Changes to data ownership or usage rights
• Whether customer data may be used to train, fine-tune, evaluate, or improve AI models
• Any opt-out, consent, retention, anonymization, or security requirements
• Whether the provider may share customer data with affiliates, subprocessors, or third parties

AI Training and Customer Data

No conclusion can be drawn about AI-model training from the information provided. The statement that approximately 4,172 words were added does not reveal whether the new language:

• Permits use of customer content, prompts, outputs, or usage data for model training
• Applies such use by default or only with customer consent
• Distinguishes between personal data, confidential information, and de-identified data
• Allows human review or vendor access to customer data
• Gives the customer an effective opt-out right
• Limits training to aggregated or anonymized information
• Allows use of data after termination or for an unlimited period

Risk Assessment

Assessment: Unable to determine. The addition of a large volume of text may materially change the agreement, but the legal and commercial impact cannot be evaluated without the underlying redline.

Information Needed

Please provide the actual diff, including the text marked with:

• {additions}
• [deletions]
• []{replacements}

If the document is too large, provide at least the provisions concerning:

1. Customer data and content

2. Confidentiality and privacy

3. Artificial intelligence, machine learning, or model training

4. Data retention and deletion

5. Subprocessors and third-party providers

6. Intellectual property and license grants

7. Security, liability, and indemnification

Key Changes and Risks

1. Terms expanded from website use to Datadog’s broader product ecosystem

• The prior terms applied to Datadog’s “Site” and Site content.
• The revised text appears to reference Datadog’s extensive product and platform offerings, including monitoring, security, developer tools, AI, automation, and digital-experience products.
• This may create uncertainty about whether the Website Terms apply to product features accessed through the website, including AI-enabled features, or whether those products remain governed exclusively by the Master Subscription Agreement (MSA) or another written contract.

Risk: Customers may have difficulty determining which contract controls if the Website Terms, MSA, product terms, and other online policies overlap or conflict.

2. Express treatment of AI services and AI-enabled features

The revised text adds or references:

• “AI” services and capabilities;
• AI agents, investigation tools, security analysts, workflow automation, and related features; and
• a provision stating that use of the Datadog Service—including through a free trial—is subject to the MSA or another separately agreed written contract.

It also appears to add language providing that use of certain AI services or features may be subject to the MSA or another separate contract.

Risk: AI functionality may be brought within contractual terms that were not previously prominent in the Website Terms. Customers should confirm whether the MSA contains separate provisions on AI inputs, outputs, model providers, retention, confidentiality, and permitted data use.

3. No clear new authorization to train AI models identified

The diff does **not clearly add an express provision authorizing Datadog to use customer data, customer content, prompts, logs, telemetry, or outputs to train, fine-tune, or improve AI models.

The existing “Your Content” provision remains important. It grants Datadog a broad, perpetual, worldwide, irrevocable, royalty-free, sublicensable license to use posted or made-available content, including to create derivative works, distribute, sell, and otherwise use it without restriction. However, it expressly excludes personal information subject to the Privacy Policy.

Risk: Although this is not a newly identified AI-training clause, its breadth could potentially support model-improvement uses unless limited elsewhere by the MSA, Privacy Policy, product-specific terms, or a data-processing agreement. The documents should be reviewed together for an explicit “no training” commitment or opt-out.

4. Scope and drafting concerns

• The diff contains extensive inserted navigation and product-marketing text, making the operative legal language difficult to identify.
• References to products and AI capabilities may be descriptive rather than binding contractual language.
• The updated terms appear to preserve Datadog’s ability to modify terms by posting a revised version online.

Recommended review

Confirm the controlling agreement for each AI feature and specifically verify:

1. whether customer data is used for training or improving models;

2. whether data is shared with third-party model providers;

3. whether inputs and outputs are retained;

4. confidentiality and security protections; and

5. available opt-outs or contractual “no training” restrictions.

Key Changes and Risks

1. Expanded scope of the “Site” and Datadog entities

• The Terms now expressly cover the website and “any related websites.”
• Datadog, Inc. and its affiliates are collectively defined as “Datadog.”
• The Terms apply to accessing or using the Site and its content, rather than only the previously described website/platform language.

Risk: Users may be bound by the Terms when visiting related Datadog websites or using additional Site features operated by affiliates, even where the affiliate is not separately identified.

2. Expanded references to Datadog products and AI features

The revised language references a substantially broader set of Datadog products and capabilities, including:

• AI Agents, Bits AI Agents, Bits Chat, Bits Code, Bits Investigation, Bits Security Analyst, and Agent Builder;
• AI Integrations and Agent Observability;
• MCP Server, Pup CLI, Agent Directory, Watchdog, and other automation or monitoring capabilities.

Risk: The broad product references may make it less clear which contractual terms apply to a particular product, especially where features combine Site functionality with hosted Datadog services.

3. Hosted services are expressly governed by other agreements

The Terms continue to state that the hosted application and infrastructure-monitoring services are governed by Datadog’s Master Subscription Agreement (“MSA”), or another written contract, rather than these Website Terms.

The update expressly extends this concept to the Service, including access through a free trial. AI features may also be governed by the MSA, subscription agreement, or another separately negotiated contract.

Risk: Customers may need to review multiple documents to determine the applicable rights, liability limits, data-use provisions, and AI terms. The referenced MSA may contain materially different provisions from these Website Terms.

4. Customer content license remains broad

The “Your Content” provision grants Datadog a perpetual, worldwide, irrevocable, royalty-free, sublicensable license to use posted or uploaded content, including to reproduce, modify, distribute, sell, and otherwise use it without restriction.

The provision excludes personal information subject to the Privacy Policy, but it does not expressly exclude confidential business information, source code, prompts, telemetry, logs, or AI inputs.

Risk: If customer information is submitted through the Site rather than a contracted Service, the license could be interpreted broadly. Customers should avoid submitting confidential or regulated data unless the applicable agreement clearly protects it.

5. AI-model training and use of customer data

• The supplied diff does not expressly add language authorizing Datadog to train AI models using customer data.
• It also does not expressly prohibit such training.
• The new AI product references could cause AI-related data-use terms in the MSA, product terms, privacy policy, or other incorporated agreements to become relevant.
• The broad “Your Content” license could create uncertainty if customer prompts, outputs, code, logs, or other inputs are treated as Site Content.

Recommended review: Confirm in the governing MSA or AI-specific terms whether customer data, prompts, inputs, outputs, telemetry, or de-identified data may be used to train, fine-tune, evaluate, or improve Datadog or third-party AI models, and whether opt-out, retention, deletion, confidentiality, and subprocessors protections apply.

Summary of Important Changes

1. Scope and document structure

• The opening language appears to be substantially replaced with a broad “Website Terms” heading and references to Datadog’s products, platform capabilities, monitoring, security, digital experience, software delivery, and other services.
• Much of the added text appears to be website navigation, product listings, marketing content, and footer material rather than operative contractual language. This creates uncertainty about which portions are intended to form part of the legal terms.
• The prior terms clearly stated that they governed use of the Site and its content, while excluding Datadog’s hosted application and infrastructure-monitoring services.

2. Hosted services and AI services

• The prior exclusion for Datadog’s hosted application and infrastructure-monitoring services is being expanded or supplemented with references to numerous Datadog offerings, including:
• AI-powered monitoring and observability;
• AI agents and related “Bits” products;
• security, application, infrastructure, and digital-experience products;
• software-delivery and developer tools.
• The revised language appears to state that use of the Service—even through a free trial—is subject to Datadog’s Master Subscription Agreement or another separately agreed written contract.
• This is important because the Website Terms may not govern customer data processed through the hosted platform. The applicable MSA, product terms, AI terms, data-processing agreement, and privacy documentation should therefore be reviewed separately.

3. AI-model training and customer data

• No express provision authorizing or prohibiting the use of customer data to train AI models is visible in the supplied diff.
• The additions refer to AI products, AI agents, AI-powered monitoring, and model performance, but do not clearly state:
• whether customer data, prompts, logs, telemetry, code, or outputs may be used for model training;
• whether data is used to train Datadog’s models or third-party models;
• whether training is for product improvement, generalized models, or customer-specific models;
• whether customers can opt out;
• whether data is anonymized, aggregated, retained, or deleted; or
• whether Datadog may use customer data to evaluate or improve AI features.
• This absence is itself a risk: the governing AI/data-use terms may be located in the MSA or separate product documentation, and the revised Website Terms do not clarify the allocation of rights.

4. Customer and legal risks

• Customers could mistakenly assume that accepting the Website Terms authorizes use of the full Datadog platform, while the operative terms may be elsewhere.
• Expanded product references may broaden the apparent scope of services without adding corresponding data-protection, confidentiality, security, or AI-use limitations.
• Customers should confirm which agreement controls in the event of conflict and obtain a clear written statement on AI training, opt-out rights, data retention, subprocessors, and use of customer content.

Summary of Changes

Overall Change

• The diff states that approximately 3,544 words were removed, but it does not identify the deleted language.
• Because the actual deletions are not provided, it is not possible to determine precisely which contractual rights, obligations, limitations, or protections changed.

Customer Data and AI Training

• No specific AI-training language is shown in the provided diff.
• Therefore, it cannot be determined whether the revision:
• Permits or prohibits using customer data to train, fine-tune, or improve AI models;
• Changes whether customer data may be used in aggregated, de-identified, or pseudonymized form;
• Removes consent, notice, opt-out, or deletion rights;
• Changes ownership or licensing rights in customer data, inputs, outputs, or derived data;
• Expands access to customer data by affiliates, vendors, or AI providers; or
• Alters security, confidentiality, retention, or data-use restrictions applicable to AI systems.

Potential Legal Risks from the Deletions

The removal of a large block of text could materially affect the customer if it deleted any of the following:

• Restrictions on secondary use, including use for analytics, product improvement, or AI model training;
• Confidentiality and data-protection commitments;
• Data ownership and license limitations;
• Representations, warranties, or indemnities relating to privacy, security, or intellectual property;
• Deletion, return, or retention requirements after termination;
• Audit, transparency, or compliance rights;
• Limits on subcontractors or third-party model providers; or
• Liability caps, exclusions, or remedies applicable to data misuse.

Assessment Limitation

This diff is insufficient for a reliable legal comparison. The deleted 3,544 words—or at least the sections concerning data use, artificial intelligence, confidentiality, privacy, security, intellectual property, and termination—should be provided before concluding whether the customer’s data may now be used to train AI models or whether customer protections have been weakened.

Structured Summary of Important Changes

1. Scope of the Privacy Policy Expanded

The policy now applies to Datadog:

• Products and services;
• Events;
• Websites; and
• Applications that link to the policy.

These are collectively defined as the “Datadog Products.”

Risk / significance

Adding applications expressly broadens the policy’s coverage beyond websites and may bring additional Datadog applications, interfaces, and product experiences within the same data-collection and usage permissions. Customers should confirm which applications link to the policy and whether separate product-specific notices or contractual terms apply.

2. Additional Purpose: Enhancing User Experiences

The revised introductory language adds that Datadog uses personal information in connection with applications that “link to this Policy” and to enhance user experiences.

Risk / significance

“Enhance user experiences” is broad and undefined. It could support activities such as personalization, product optimization, interface improvements, recommendations, experimentation, or analytics. The policy does not limit this purpose to aggregated, de-identified, or non-customer data.

3. Customer Data and AI Model Training

No express AI-training authorization identified

The diff does not add language expressly stating that Datadog may use customer data, Product information, logs, traces, metrics, code, prompts, outputs, or other submitted content to:

• Train artificial-intelligence or machine-learning models;
• Fine-tune or improve foundation models;
• Create datasets for AI training; or
• Share customer data with AI-model providers for those purposes.

The revised policy continues to state that Datadog may use collected information, including Product information, for research and to improve or create Datadog Products. However, that language is general and is not specifically framed as AI training.

Residual risk

Because “Product information” may include information submitted through Datadog Products—such as logs, traces, and metrics—and because “improve and create new Datadog Products” is broad, the policy may still provide a general basis for analytics, product development, or potentially AI-related development unless customer agreements, data-processing terms, or product-specific documentation impose stricter limits.

Customers should seek confirmation of:

• Whether customer content is used to train or fine-tune AI models;
• Whether data is used only for service delivery or also for product improvement;
• Whether data is aggregated or de-identified first;
• Whether customer opt-out controls exist; and
• Whether AI vendors or subprocessors receive customer data.

4. Website Navigation and Product Listings

The diff adds or updates extensive Datadog product and navigation listings, including AI products and services such as AI Agents, Agent Builder, Bits Chat, Bits Code, and related tools.

Risk / significance

These appear primarily to be website-content updates rather than substantive privacy permissions. Their inclusion does not, by itself, authorize AI training or materially change data-use rights.

5. Overall Assessment

The principal substantive changes are the expanded coverage of linked applications and the addition of the broad “enhance user experiences” purpose. No clear new contractual right to train AI models on customer data appears in the provided diff, but the general product-improvement language remains broad enough to warrant clarification.

Key Changes and Legal Risks

1. Terms expanded from website use to Datadog products and platform

The opening language is substantially broadened. It no longer focuses only on accessing Datadog’s websites and website content. It now refers to Datadog’s integrated platform and a wide range of monitoring, security, observability, developer, digital experience, service management, and AI capabilities.

Risks and implications:

• The terms may now purport to apply to use of Datadog products and platform features, not merely the public website.
• The scope of covered services is very broad and includes features that may process operational, security, application, code, log, session, and other potentially sensitive data.
• It may be unclear whether these Website Terms supplement, replace, or merely point to the Master Subscription Agreement (“MSA”) for product use.

2. AI products and capabilities are expressly identified

The revised material adds or references numerous AI-related products and capabilities, including:

• AI Monitoring and AI Observability
• Bits AI Agents, Bits Chat, Bits Agent Builder, Bits Code, Bits Investigation, and Bits Security Analyst
• Watchdog and Agent Observability
• AI Integrations
• AI Impact
• Agentic and embedded features

The terms also state that use of the Service is subject to Datadog’s MSA or another separately agreed contract, including use through a free trial. The revised text appears to add or emphasize AI services as part of the Service.

Risks and implications:

• Customers may become subject to additional contractual terms simply by using AI functionality, including during a free trial.
• The incorporated MSA or other contract may contain the operative rules on data processing, confidentiality, intellectual property, AI outputs, and model development.
• Customers should confirm which agreement controls if these Website Terms conflict with the MSA, product terms, order form, or AI-specific terms.

3. No express customer-data training authorization identified

The diff does not clearly add language stating that Datadog may use customer data, customer content, prompts, logs, code, telemetry, or AI inputs/outputs to train, fine-tune, improve, or evaluate AI models.

Likewise, it does not clearly add:

• A right to use customer data for generalized model training;
• A right to retain customer data for AI development;
• A customer opt-out or consent mechanism; or
• A restriction that customer data will not be used for model training.

However, the expanded references to AI services make this issue more important. The controlling MSA, product-specific terms, privacy documentation, or AI terms should be reviewed separately.

4. Substantial website-content additions appear non-operative

The diff includes a very large amount of navigation, product-listing, marketing, and footer content. These additions appear primarily descriptive or website-generated rather than contractual. Nevertheless, listing products may broaden ambiguity about which services are covered by the terms.

Recommended review points

• Obtain and review the current MSA and any AI-specific terms.
• Confirm whether customer data is used to train or improve Datadog’s or third-party AI models.
• Check data-retention, subprocessors, confidentiality, and opt-out provisions.
• Clarify precedence among these Website Terms, the MSA, order forms, and product documentation.

Summary

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

> “Added approximately 4172 words to the document”

Accordingly, it is not possible to identify:

• New or modified legal obligations
• Changes to liability, indemnities, warranties, termination, or governing law
• Changes to confidentiality or data-protection terms
• Whether customer data may be used to train, fine-tune, evaluate, or improve AI models
• Whether data may be shared with affiliates, subprocessors, or third-party model providers
• Any opt-out, consent, deletion, retention, or security provisions relating to AI training

AI Training and Data-Use Risk

No conclusion can be drawn about AI-model training from the information supplied. The added 4,172 words could contain significant permissions or restrictions, but the relevant wording is not shown.

When reviewing the full addition, particular attention should be paid to clauses stating that the provider may:

• Use customer content, prompts, inputs, outputs, usage data, or metadata to “improve,” “develop,” “train,” or “fine-tune” models
• Use data in aggregated, anonymized, de-identified, or pseudonymized form
• Retain data after termination or deletion requests
• Share data with affiliates, contractors, subprocessors, or third-party AI providers
• Combine customer data with other users’ data
• Use human reviewers or automated systems to evaluate customer content
• Claim ownership or broad license rights over customer inputs or outputs
• Exclude certain services, plans, or API products from an opt-out or confidentiality commitment

Information Needed

Please provide the actual redlined text, including the 4,172 added words and any surrounding provisions. The analysis should compare the new language against the prior version and identify the practical impact and risk level of each material change.

Summary of Changes

Scope of the Diff

The diff states that approximately 3,517 words were removed, but it does not identify which provisions were deleted or provide the deleted text. As a result, the specific legal and commercial effects cannot be determined reliably.

Potentially Important Risks

The deletion of a substantial amount of language may have removed provisions relating to:

• Customer data rights and permitted uses
• Confidentiality and data-security obligations
• Use of data to train, fine-tune, or improve AI models
• Ownership of inputs, outputs, and derivative materials
• Data retention and deletion
• Service limitations, warranties, and liability
• Indemnification
• Audit, compliance, or regulatory obligations
• Termination rights and post-termination data handling

Removing limitations or safeguards could materially increase the customer’s exposure, particularly if the remaining agreement grants broad rights to use customer content or reduces the provider’s obligations.

AI-Training Provisions

The available diff contains no identifiable language addressing AI-model training. Therefore, it is not possible to determine whether the agreement now:

• Permits or prohibits training models on customer data;
• Allows use of customer data for model improvement, analytics, or product development;
• Distinguishes between customer inputs, outputs, prompts, usage data, and anonymized or aggregated data;
• Requires customer consent or provides an opt-out;
• Applies different rules to personal, confidential, or regulated data;
• Requires deletion of training data after termination; or
• Protects the customer from having its data used to benefit other customers.

The absence of visible AI-training language is not evidence that such use is prohibited. Any surviving general license, data-use, confidentiality, or service-improvement provision could still govern the issue.

Recommended Review

Obtain the full before-and-after versions, or at least the text of the 3,517 deleted words. The review should specifically compare:

1. Customer data-use and license grants;

2. AI training and model-improvement rights;

3. Definitions of “Customer Data,” “Input,” “Output,” and “Usage Data”;

4. Confidentiality and security commitments;

5. Retention, deletion, and termination provisions; and

6. Liability allocation for unauthorized data use or disclosure.

On the current information, the principal conclusion is that a substantial and potentially material set of protections or restrictions may have been removed, but the exact changes and AI-related risks cannot be confirmed.

Summary

The provided diff does not include the text of the amendments. It only states:

> “Added approximately 3518 words to the document”

Accordingly, it is not possible to determine:

• Which contractual provisions were added or changed;
• Whether obligations, liabilities, fees, warranties, indemnities, or termination rights were modified;
• Whether the customer’s data may be accessed, retained, disclosed, sold, or shared with third parties;
• Whether customer data, prompts, outputs, or usage information may be used to train, fine-tune, evaluate, or improve AI models;
• Whether the customer can opt out of AI training or require deletion of data;
• Whether confidential information or personal data receives additional protections;
• Whether the provider obtains broader rights to use derived data, aggregated data, telemetry, or de-identified data; or
• Whether new security, privacy, compliance, or audit obligations have been introduced.

AI-training risk

No conclusion can be reached about AI-model training because the actual added language is not provided. The statement that approximately 3,518 words were added does not identify whether any of those words concern AI, machine learning, model improvement, data usage, or confidentiality.

Required information

Please provide the full redlined text, including:

• Additions in {};
• Deletions in []; and
• Replacements in []{}.

Without the substantive diff, any legal-risk assessment—particularly regarding the use of customer data to train AI models—would be speculative.

Summary of Important Changes

1. Broader scope of the Terms

• The Terms now apply to the Datadog website and related websites collectively referred to as the “Site.”
• Datadog, Inc. and its affiliates are collectively defined as “Datadog.”
• The opening language is expanded from use of the website to accessing or using the Site, its content, and a much broader range of Datadog products and capabilities.
• Users are expressly bound by the Terms when accessing or using the Site, including through free trials.

Risk: The Terms may now apply to a wider range of users, products, and interactions than before. Website users could unintentionally accept contractual obligations applicable to product-related or AI-related functionality.

2. New AI and agent-related services

The revised language expressly references Datadog’s AI offerings, including:

• AI Agents
• Bits Chat, Bits Agent Builder, Bits Code, and Bits Investigation
• Security Analyst
• Agent Observability
• AI Integrations
• Watchdog and other AI-enabled capabilities

Use of the Service, including AI features accessed through a free trial or through the Datadog platform, is now addressed through Datadog’s Master Subscription Agreement (MSA) or another applicable written agreement.

Certain AI features may also be governed by the MSA or a separate agreement, including contracts incorporated by reference or separately agreed between the customer and Datadog.

Risk: The governing terms for AI functionality may not be contained entirely in these Website Terms. Customers may need to review the MSA and any product-specific agreements to understand data handling, confidentiality, intellectual property, security, liability, and AI-use provisions.

3. Customer data and AI-model training

The diff does not clearly add an express provision stating that Datadog may use customer data to train, fine-tune, or improve AI models. It also does not clearly prohibit such use.

The revised text appears to expand references to AI services and Datadog’s platform, but the supplied diff does not show a specific customer-data license or training authorization connected to AI models.

Key implication: The absence of a visible training clause does not resolve the issue. Any AI-data rights may instead appear in:

• The MSA;
• Product-specific terms;
• A data-processing agreement;
• Privacy or security documentation; or
• Separate AI feature terms.

Customers should confirm whether “Customer Data,” telemetry, prompts, outputs, logs, or other submitted content may be used for model training or service improvement, and whether opt-out rights exist.

4. Expanded contractual representations

Users entering the Terms for an organization now expressly represent that they have authority to bind that entity, and “you” includes both the individual and the entity.

Risk: An employee or contractor could bind the organization by accessing the Site or AI-enabled features without clear internal authorization.

5. Overall assessment

The principal change is a substantial expansion of the Terms to cover Datadog’s broader product ecosystem and AI capabilities. The diff creates potential uncertainty—not a clear affirmative grant—regarding use of customer data for AI training. The MSA and any AI-specific terms should be reviewed before using AI features with confidential, personal, regulated, or proprietary data.

Summary of Important Changes

1. Policy date and covered services

• The policy appears to be updated to March 12, 2026.
• The description of covered offerings is reorganized and expanded to include Datadog’s broader product portfolio, including monitoring, security, software delivery, digital experience, AI, and related applications.
• The policy now expressly covers “applications that link to this Policy”, in addition to Datadog products, services, events, and websites. This may expand the policy’s scope to additional Datadog applications.

2. New or clarified customer-data uses

The revised text states that Datadog collects “Product information” submitted through Datadog Products, including information contained in:

• Logs
• Traces
• Metrics
• Potentially names, email addresses, IP addresses, online identifiers, and mailing addresses contained in that data

Datadog may use collected information, including Product information:

• To provide the services
• For research
• To monitor and analyze trends, usage, and other activities
• To improve existing products
• To create new products
• To combine information with data obtained from other sources to understand customer needs and develop services

AI-model training

• The diff does not expressly add or authorize training AI models, nor does it mention “training,” “machine learning,” “large language models,” or use of customer data to develop AI models.
• However, the broad language allowing Product information to be used for research, product improvement, and creation of new products could potentially encompass AI-related development unless restricted elsewhere in the applicable customer agreement, data-processing addendum, or product terms.
• The text does not provide a clear customer opt-out specifically for AI development or model training. Customers should therefore check Datadog’s service terms, AI-specific terms, order form, and data-processing agreement for more precise restrictions.

3. Broader data-sharing language

The policy retains broad sharing permissions for:

• Service providers, including hosting, authentication, cybersecurity, anti-fraud, advertising, analytics, and search providers
• Affiliates, including current and future related companies
• Business transfers, such as mergers, acquisitions, financing, or asset sales
• Legal, security, and compliance purposes

This creates potential exposure to additional processors or affiliates, although the policy does not specifically state that customer Product information will be shared for AI training.

4. Expanded privacy-rights description

The revised policy more clearly describes rights and controls, including:

• Access, correction, deletion, portability, restriction, and objection rights
• Withdrawal of consent
• U.S. opt-outs for targeted advertising, automated profiling, and sale or sharing of personal information
• Privacy preference signals such as Global Privacy Control
• Appeal rights for denied requests

These rights may apply differently depending on the individual’s jurisdiction and may not override contractual customer-account terms or retention obligations.

5. Key customer risks

• Ambiguous scope: “Product information” may include sensitive or personal data placed in logs, traces, and metrics.
• Broad improvement language: It is not limited to aggregated, de-identified, or service-generated data.
• No explicit AI-training limitation: The policy neither confirms nor rules out use of customer data for AI development.
• Customers should seek a written commitment that customer content will not be used to train generalized AI models, unless expressly authorized.

Structured Summary of Important Changes

1. Scope expanded beyond a website

• The terms are retitled “Website Terms of Use” and now appear connected to a broader Datadog platform and product ecosystem.
• The revised text refers to accessing or using Datadog’s platform, products, monitoring, security, digital experience, software delivery, and other capabilities, rather than only visiting a website or viewing website content.
• The change may unintentionally broaden the circumstances in which these terms apply, potentially covering interactions with product pages, integrated tools, and platform features.

2. AI services and features are expressly referenced

• The revised material adds extensive references to Datadog’s AI-related offerings, including AI Agents, Bits, Watchdog, Agent Observability, AI Integrations, AI Impact, and other AI capabilities.
• It also adds language indicating that certain AI services may improve model performance, identify root causes, detect anomalies, or provide automated assistance.
• Customers using these services may therefore be subject to additional terms, including the Master Subscription Agreement or another written contract, even when access occurs through a free trial.

3. Customer data and AI-model training

• No clear express provision was identified stating that customer data will be used to train Datadog’s AI models.
• The diff does not appear to add a specific authorization for Datadog to:
• use customer data to train, fine-tune, or improve general-purpose AI models;
• retain prompts, inputs, outputs, logs, or telemetry for model training;
• share customer data with AI providers for training; or
• exclude customer data from training.
• However, the expanded references to AI services create ambiguity about how customer content, prompts, telemetry, logs, and outputs may be handled. The governing Master Subscription Agreement, privacy terms, product-specific terms, or AI-specific documentation should be reviewed for the operative data-use rules.
• Customers should seek written confirmation that their data will not be used for model training except as expressly authorized, and should verify retention, subprocessors, anonymization, and deletion provisions.

4. Contract hierarchy and incorporation risk

• The revised text more prominently directs users of Datadog’s hosted services and AI features to the Master Subscription Agreement or another separately agreed contract.
• This creates a risk that important data-processing, confidentiality, security, and AI-use terms are located outside these website terms and may differ by product or account.
• Customers should confirm which agreement controls in the event of conflict and whether product-specific or online terms can be changed unilaterally.

5. Drafting and implementation concerns

• The diff contains substantial apparent website-navigation and product-catalogue text inserted into legal provisions. This may be a formatting or publishing error.
• If published as written, the text could create uncertainty about definitions, contractual scope, and enforceability. Datadog should clarify or correct the consolidated version before reliance.

Structured Summary of Important Changes

1. Expanded scope of the Terms and Datadog entities

• The operator is now defined as Datadog, Inc. or one of its affiliates, collectively “Datadog.”
• The “Site” is expanded to include the website and related websites.
• The Terms apply to accessing or using the Site and its content, rather than merely referring to a website.

Risk: Users may be contracting with, or granting rights to, a broader group of Datadog affiliates. The definition of “you” also continues to include both the individual user and any entity the user represents, creating potential personal and corporate responsibility.

2. Greater separation between the Site and Datadog services

• The Terms continue to state that they do not govern Datadog’s hosted application and infrastructure-monitoring service.
• The exclusion is broadened to cover the Service and “other services.”
• The revised language expressly refers to accessing or using the Service, including through a free trial, and directs users to the Master Subscription Agreement (MSA) or another written contract.

Risk: Users of Datadog’s products—including newer AI-related products—may be governed primarily by the MSA or another contract rather than these Website Terms. The applicable agreement may therefore contain different data-use, confidentiality, liability, and AI provisions.

3. AI products and AI-enabled services

• The revised material adds or identifies numerous AI-related products and capabilities, including AI Agents, Bits products, Agent Observability, AI Integrations, Watchdog, and related platform capabilities.
• Use of the relevant Service is stated to be subject to the MSA or another separately agreed contract.

Risk: AI functionality appears to be brought within the Service framework, but the diff does not clearly state which AI features are covered by which agreement. Customers should confirm whether their order form, MSA, product terms, or AI-specific terms control.

4. Customer data and AI-model training

• No express new provision authorizing Datadog to use customer data to train AI models is visible in this diff.
• The existing “Your Content” provision grants Datadog a broad, perpetual, worldwide, sublicensable license to use content posted on the Site, including to reproduce, modify, distribute, sell, and otherwise use it. However, it expressly excludes personal information governed by the Privacy Policy.
• That provision concerns Site content and does not clearly authorize training models using customer data submitted to Datadog’s hosted Services.

Key uncertainty/risk: The revised Website Terms do not resolve whether customer data, telemetry, logs, prompts, outputs, or other Service data may be used to train Datadog’s or third-party AI models. That issue may instead be addressed in the MSA, product-specific terms, privacy documentation, or an AI addendum. Customers should obtain an explicit written commitment regarding whether their data is:

• used for model training or fine-tuning;
• used only to provide or secure the Service;
• shared with AI subprocessors; and
• retained for training after termination or deletion requests.

5. Practical review priority

Customers using Datadog AI features should review the MSA and applicable product/AI terms together with these Website Terms. The main unresolved issue is not an apparent new training right in this diff, but the possibility that AI data-use rights are located in a different incorporated agreement.

Key Changes and Legal Risks

1. Terms appear to be repurposed for Datadog products and services

The prior language governed access to Datadog’s websites and website content. The revised text appears to expand or redirect the terms toward Datadog’s broader product, platform, monitoring, security, software-delivery, and AI offerings.

Risk: The scope is unclear because the diff contains extensive product-navigation and website text, apparently inserted into operative language. If this text is intended to be contractual, it may create ambiguity about which products and services are covered.

2. AI services and capabilities are expressly referenced

The revised material adds references to:

• AI agents and embedded AI features;
• Bits AI Agents, Bits Chat, Bits Code, Bits Investigation, and Security Analyst;
• Watchdog AI and AI integrations;
• Agent Observability and related AI capabilities.

It also appears to state that certain AI services are subject to Datadog’s Master Subscription Agreement (MSA) or another separately negotiated written contract, including where the service is accessed through a free trial.

Risk: Customers may be bound by additional terms that are not included in these Website Terms. The documents should clearly identify which agreement controls if the Website Terms, MSA, product terms, or AI-specific terms conflict.

3. No express customer-data AI-training provision identified

The diff does not clearly add language stating that Datadog may use customer data, customer content, telemetry, logs, prompts, outputs, or other service data to train, fine-tune, evaluate, or improve AI models.

Likewise, it does not clearly add a prohibition on using customer data for AI training, nor does it specify:

• whether customer data is used to train Datadog’s models;
• whether data is shared with AI subprocessors or model providers;
• whether data is de-identified or aggregated first;
• whether customers can opt out;
• retention, deletion, or model-memory obligations; or
• ownership and permitted use of AI inputs and outputs.

Important: These issues may instead be addressed in the MSA, product-specific terms, privacy documentation, or AI terms incorporated by reference. Those documents should be reviewed before accepting the revised terms.

4. Existing broad license for user-posted content remains significant

The prior “Your Content” provision grants Datadog a perpetual, worldwide, irrevocable, royalty-free, sublicensable license to use posted content broadly. It excludes personal information governed by the Privacy Policy, but the exclusion may not cover business data, telemetry, logs, prompts, or other information submitted through Datadog products.

Risk: If this provision applies to product or AI inputs, it could support broad secondary use unless limited by the MSA or another governing agreement.

5. Practical recommendations

• Confirm whether the revised text is genuinely contractual or contains website navigation/content artifacts.
• Obtain and compare the current MSA and any AI-specific terms.
• Require explicit language addressing AI training and customer-data use.
• Clarify agreement priority, data ownership, confidentiality, subprocessors, opt-out rights, and deletion obligations.

Structured Summary of Important Changes

1. Privacy Policy Replaced/Restored as Primary Content

The diff appears to replace a highly fragmented product-navigation page with a substantially expanded Datadog Privacy Policy, dated March 12, 2026. The revised policy applies to Datadog products, services, events, websites, and applications that link to it.

Risk: The policy’s scope is broad and may cover information processed through many newly listed products, including AI, security, monitoring, developer, and workflow products. Customers should confirm which products and data flows are governed by the policy versus separate contracts, product terms, or data-processing agreements.

2. Expanded Definition of Collected “Product Information”

The revised policy expressly states that Datadog collects personal information submitted through Datadog Products, including information in:

• Logs
• Traces
• Metrics

Examples include names, email addresses, IP addresses, online identifiers, and mailing addresses.

Risk: Customer telemetry may contain personal, confidential, regulated, or sensitive information. The policy does not appear to limit collection to data intentionally supplied as personal information; data embedded in customer logs or monitoring records could be included.

3. Use for Research, Improvement, and New Products

A new or clarified use purpose permits Datadog to use information collected—including Product information—to:

• Research and analyze trends and usage;
• Improve existing Datadog Products; and
• Create new products and services.

Datadog may also combine information about individuals with information obtained from other sources.

AI-training implications: The diff does not expressly state that customer data will be used to train artificial-intelligence or machine-learning models. However, the broad authorization to use Product information to improve and create products could potentially encompass development, evaluation, or refinement of AI features, particularly given the expanded references to AI Agents, AI Impact, Bits, Watchdog, Agent Observability, and related products. The policy does not provide clear limits, de-identification requirements, customer opt-out, model-retention rules, or a commitment that customer content will not be used for model training.

4. Expanded Marketing and Advertising Uses

The policy authorizes personalized advertising, behavioral or targeted advertising, campaign measurement, conversion attribution, and sharing with analytics, advertising, and search-engine providers. U.S. notices describe categories of information shared for cross-context behavioral advertising.

Risk: Customer or user identifiers, usage data, geolocation, inferences, and electronic activity may be used or disclosed for advertising purposes, subject to applicable choices and opt-outs.

5. New or Expanded Privacy Rights

The policy details rights under GDPR, UK GDPR, Swiss law, and U.S. state laws, including access, correction, deletion, restriction, objection, portability, opt-out of targeted advertising, automated profiling, sale/sharing, and limits on sensitive-information use.

6. International Transfers and Data Retention

The policy confirms international transfers, including to the United States, using mechanisms such as Standard Contractual Clauses and the Data Privacy Framework. Retention periods are determined by contractual terms, legitimate interests, legal obligations, and data characteristics; anonymized information may be retained and used without further notice.

Overall risk: The revised policy materially broadens and clarifies Datadog’s potential uses of telemetry and personal information, but does not clearly resolve whether customer data may be used to train AI models. This should be addressed in the customer agreement, DPA, or product-specific terms.

Executive Summary

The diff appears to update the Website Terms of Use to cover Datadog’s broader product ecosystem, including AI products and agents. It also clarifies that use of Datadog’s hosted services—including through a free trial—is governed by the Master Subscription Agreement (MSA), rather than these Website Terms.

The provided diff is highly textually corrupted, so some wording and scope should be verified against the clean revised Terms.

Important Changes and Risks

1. Broader definition of Datadog and the Site

• Added: The Site may be owned and operated by Datadog, Inc. or one of its affiliates.
• Added/clarified: The “Site” includes the Datadog website and related websites.

Risk: Users may be contracting with, or granting rights to, a Datadog affiliate rather than Datadog, Inc. alone. This could affect enforcement, applicable entities, and allocation of responsibility.

2. Expanded coverage of Datadog services

• The prior exclusion covered the hosted application and infrastructure-monitoring service and “other services.”
• The revised language appears to define a broader “Service” and expressly includes Datadog’s extensive product portfolio, including AI-related services.

Risk: More products may be brought within the MSA and its terms, even where users access them through the Site or a free trial. Customers should confirm which agreement governs each product and whether product-specific terms also apply.

3. New AI Service language

• Added: A Datadog AI Service, apparently including AI agents, AI integrations, Bits products, and related capabilities.
• Added: If a user accesses or uses the AI Service—even through a free trial—the user is subject to the MSA or another separately agreed contract.

Risk: AI functionality may be subject to contractual terms that are not displayed in the Website Terms. Customers should review the MSA and any AI-specific documentation for confidentiality, security, output ownership, acceptable-use, liability, and data-processing provisions.

4. Customer data used to train AI models

• No express change is visible in this diff stating that customer data will—or will not—be used to train Datadog’s AI models.
• The additions identify and expand AI services, but do not clearly address:
• whether Customer Data is used for model training or fine-tuning;
• whether data is used to improve Datadog or third-party models;
• whether prompts, outputs, telemetry, or feedback are retained;
• opt-out rights or enterprise restrictions; or
• use of de-identified or aggregated data.

Risk: The absence of a training restriction in these Terms does not establish that training is prohibited. The governing MSA, privacy notice, product terms, or AI documentation may contain the relevant permission or limitation. This should be specifically confirmed before submitting confidential, personal, regulated, or proprietary data to AI features.

5. Free-trial users

• Clarified: AI Service users are subject to the MSA even when using a free trial.

Risk: Trial use may create binding obligations and expose customer data to the same AI processing terms as paid use.

Structured Summary of Important Changes

> Caution: The diff appears heavily corrupted by webpage-navigation text being inserted into legal provisions. The conclusions below should be verified against clean versions of the old and new documents before relying on them.

1. Scope appears to shift from the website to Datadog’s broader platform

• The prior language governed the Datadog website and its content.
• The replacement text appears to refer to Datadog’s integrated platform, including monitoring, observability, security, digital experience, software delivery, service management, and AI capabilities.
• This could materially expand the circumstances in which the terms apply, potentially reaching use of product interfaces and platform features rather than merely browsing the website.

2. AI products and agents are expressly identified

The replacement text lists or references AI-related offerings, including:

• AI Agents
• Bits Chat, Bits Code, Bits Investigation, and Security Analyst
• Agent Builder
• Watchdog AI
• AI integrations
• Agent Observability
• MCP Server and related agent tools

It also appears to state that use of these services—even through a free trial—is governed by Datadog’s Master Subscription Agreement (MSA) or another separate written contract.

Risk

Customers may need to review the MSA and product-specific terms to understand AI functionality, data handling, confidentiality, intellectual-property rights, security obligations, and liability limits. The Website Terms may no longer be the controlling document for AI services.

3. Customer-content license may have been removed or relocated

The old terms contained a broad license for content posted or made available on the Site, including rights to reproduce, modify, distribute, sell, sublicense, and otherwise use that content.

That language appears in the deleted material, and no clear replacement is visible in the additions.

Risk

• If genuinely removed, Datadog may no longer rely on this Website Terms provision for a broad license to customer-posted content.
• However, the replacement may exist in the MSA, Privacy Policy, AI terms, or another product agreement.
• Customers should not assume that deletion eliminates Datadog’s rights to use product data.

4. No clear new AI-training authorization is visible

The diff does **not clearly add language stating that customer data, prompts, logs, telemetry, content, or outputs may be used to train, fine-tune, evaluate, or improve AI models.

Likewise, it does not clearly add a promise that customer data will not be used for AI training.

AI-data risk

The absence of an explicit training provision is not necessarily protective. AI data-use rights may be addressed in:

• The MSA
• AI-specific terms
• Product documentation
• The Privacy Policy
• A data-processing agreement or order form

Customers should obtain written confirmation of whether Datadog may use customer data—including service data, prompts, logs, traces, code, and AI interactions—for model training or product improvement, and whether opt-out, deletion, retention, and anonymization controls exist.

5. Expanded product references may create interpretation risk

The insertion of extensive product, feature, and navigation text makes it difficult to determine the operative legal language. It may be a website scrape rather than intentional contractual drafting.

Recommended action: Compare clean, dated versions and identify the controlling agreement for each AI-enabled service before accepting the updated terms.

Key Changes and Risks

1. Broader scope of the Terms

• The definition of the “Site” is expanded to include the Datadog website and related websites operated by Datadog or its affiliates.
• The opening language now refers broadly to using, accessing, or viewing the Site and its content, rather than only using the website.
• The Terms expressly state that accepting or using the Site legally binds the user.
• Users representing a company must confirm they have authority to bind that entity. “You” includes both the individual and the represented entity.

Risk: Employees or contractors could unintentionally bind their employer by accessing the Site or related features without clear internal authorization.

2. New and expanded AI and service references

The revised language adds or identifies numerous Datadog products and capabilities, including:

• AI Agents and AI-powered features;
• Bits Chat, Bits Code, Bits Investigation, Bits Security Analyst, Agent Builder, and MCP Server;
• Watchdog and Agent Observability;
• AI Integrations and other automation features.

The Terms now state that use of the Service—including through a free trial—may be governed by Datadog’s Master Subscription Agreement (MSA) or another written contract. AI features may also be subject to the MSA or a separate agreement.

Risk: The website Terms do not appear to provide the complete terms for AI-enabled services. Customers may need to review the MSA and any product-specific terms to determine data-use, confidentiality, security, ownership, and liability protections.

3. Customer data and AI-model training

• The visible changes do not expressly state that Datadog may use customer data to train general-purpose or product AI models.
• They also do not expressly prohibit such training.
• The existing “Your Content” provision grants Datadog a broad, perpetual, worldwide, irrevocable, royalty-free, sublicensable license to use, reproduce, modify, distribute, sell, and otherwise use content posted to the Site.
• That provision excludes personal information covered by the Privacy Policy, but it does not clearly address confidential business data, telemetry, logs, prompts, outputs, or data submitted to AI features.

Important risk: If customer material is posted or submitted through the Site, the broad content license could arguably permit uses beyond providing the service, potentially including analytics, product development, or AI-model improvement. The clause does not clearly limit use to service delivery or state whether data is used for model training.

Recommended action: Confirm in the MSA, product terms, privacy documentation, and data-processing agreement whether Datadog:

1. uses customer data to train AI models;

2. uses data only in de-identified or aggregated form;

3. retains prompts, inputs, outputs, or telemetry;

4. permits opt-out or requires consent; and

5. treats customer content as confidential.

4. Liability and enforcement

The existing Terms retain a broad indemnity, unilateral suspension/termination rights, “as-is” disclaimers, and a $200 aggregate liability cap.

Risk: These protections may be inadequate for sensitive operational, security, or confidential data processed through AI or monitoring products.

Structured Summary of Important Changes

1. Policy scope and date

• The Privacy Policy is now dated March 12, 2026.
• The scope is described more broadly as covering Datadog products, services, events, websites, and applications that link to the Policy.
• The definition of “Datadog Products” now expressly includes the broader Datadog platform and newer offerings, including AI-related products and agent features.
• The Policy adds language explaining that it describes how users can control or exercise rights concerning their personal information.

2. Customer data expressly identified as collected

The revised policy expressly identifies “Product information” submitted through Datadog Products, including information contained in:

• Logs
• Traces
• Metrics

It states that this information may include names, email addresses, IP addresses, online identifiers, and mailing addresses. This is important because customer telemetry may contain personal information, even where the customer did not intend to submit it.

The policy also says Datadog may receive information about an individual through other users’ use of Datadog Products, creating a potentially broad data-collection scope.

3. Use for research, product improvement, and new products

Datadog states that it may use information it collects—including Product information—to:

• Research and improve Datadog Products;
• Monitor and analyze trends, usage, and other activities;
• Continually improve existing products; and
• Create new products.

It may also link or combine information about an individual with information obtained from other sources.

Risk

This is a broad secondary-use authorization. It may permit use of customer-submitted logs, traces, and metrics beyond merely providing the contracted service, including for development of new Datadog capabilities. Customers should confirm whether their separate agreement, data-processing addendum, or service terms impose stricter limits on use of customer content.

4. AI-model training

The diff does not clearly add an express statement that customer data will be used to train AI models, nor does it expressly state that customer data will not be used for AI training.

However:

• The policy now covers numerous AI and agent-related products.
• The broad authorization to use Product information for research, improvement, and creation of new products could potentially encompass AI development or model improvement, depending on how Datadog interprets those purposes.
• The policy does not identify whether customer data is used for:
• Training foundation or generative AI models;
• Fine-tuning models;
• Evaluating or testing models;
• Improving prompts, agents, or automated outputs; or
• Human review of AI-related data.
• It also does not provide a customer-specific opt-out for AI training or explain whether customer data is aggregated, de-identified, or isolated before such use.

Recommended action

Customers using sensitive logs, source code, credentials, regulated data, or personal information should seek written clarification on AI training and review the governing contract and DPA for any conflicting restrictions.

5. Other notable provisions

• Datadog may use personal information for personalized marketing, advertising, analytics, and conversion attribution.
• It may share information with service providers, advertising and analytics providers, affiliates, and in business transfers.
• Retention periods remain flexible and may include anonymization, after which Datadog may use the resulting information without further notice.
• The Policy expands or restates EEA/UK/Swiss and U.S. privacy rights, including opt-outs for targeted advertising, profiling, and certain sales or sharing activities.

No

Summary of Important Changes

1. Privacy Policy substantially expanded and reorganized

The diff replaces a short introductory privacy statement with a comprehensive Privacy Policy covering:

• Categories of personal information collected
• Purposes for using personal information
• Sharing with service providers, affiliates, advertisers, and other third parties
• Security and retention
• Children’s data
• User choices and data-subject rights
• EEA/UK/Switzerland and U.S. supplemental notices
• International data transfers and Data Privacy Framework participation
• Contact and complaint procedures

Risk: The policy now contains substantially broader and more detailed descriptions of Datadog’s data practices. Customers should not assume that data submitted through Datadog is limited to basic account information.

2. Customer-submitted Product information expressly included

A new definition of “Product information” states that Datadog collects personal information submitted through use of the products, including information in:

• Logs
• Traces
• Metrics

Examples include names, email addresses, IP addresses, online identifiers, and mailing addresses.

Risk: This expressly confirms that customer telemetry and observability data may contain personal information and is within the scope of Datadog’s stated processing activities. Customers remain responsible for determining whether their logs, traces, or metrics contain regulated, confidential, or sensitive data.

3. Broad product-improvement and research use

The policy states that Datadog may use information collected—including Product information—to:

• Research, monitor, and analyze trends and usage
• Improve existing products
• Create new products
• Combine information with information obtained from other sources

It also states that Datadog will not process personal information for materially different purposes without providing an opportunity to opt out.

AI-training implications: The diff does not expressly state that customer data will be used to train, fine-tune, or develop AI models. However, the broad “research,” “improve,” and “create new products” language could potentially encompass AI-related development, particularly given the extensive addition of AI products and features. The policy does not clearly distinguish customer content from product-usage data, nor does it provide a specific customer-data opt-out from AI training.

4. Extensive AI-related products added

The revised content prominently adds AI offerings, including AI Agents, Agent Builder, Bits Chat, Bits Code, Investigation, Security Analyst, Watchdog AI, Agent Observability, and AI Integrations.

Risk: These additions signal expanded AI functionality, but the policy does not explain whether customer inputs, logs, prompts, outputs, or telemetry associated with those features are retained or used to train models.

5. Expanded third-party sharing

Datadog may share personal information with hosting, authentication, cybersecurity, anti-fraud, advertising, analytics, search-engine providers, affiliates, and entities involved in corporate transactions.

Risk: The categories are broad, and the policy does not identify specific providers or impose detailed limits on secondary use in this document.

6. New rights and transfer disclosures

The policy adds GDPR/UK GDPR rights, U.S. state privacy rights, advertising opt-outs, Global Privacy Control recognition, international-transfer mechanisms, and Data Privacy Framework disclosures.

Overall assessment: The most important commercial uncertainty is the absence of an explicit statement either permitting or prohibiting use of customer data for AI-model training. A separate service agreement, data-processing addendum, or AI-specific terms should be reviewed.

Summary of Available Changes

The supplied diff states only that approximately 3,549 words were removed. It does not identify which provisions were deleted or provide the replacement text.

Key Legal Implications

Because the deleted language is unavailable, it is not possible to determine reliably whether the amendments:

• Change the parties’ rights or obligations;
• Remove warranties, indemnities, limitations of liability, or termination rights;
• Alter confidentiality, security, privacy, or data-processing obligations;
• Change ownership or licensing of customer data;
• Add or remove audit, compliance, or regulatory protections; or
• Affect governing law, dispute resolution, or other commercial terms.

A large deletion may be significant even if no new language was added. Removing protective provisions can shift risk to the customer or leave important issues governed only by general or default legal rules.

Customer Data and AI Training

The available diff contains no specific language addressing AI, machine learning, model training, data usage, data retention, or data sharing. Accordingly, it is not possible to determine whether the revised document:

• Permits the provider to use customer data, prompts, outputs, or usage data to train or improve AI models;
• Restricts such use to de-identified, aggregated, or service-improvement purposes;
• Requires customer consent or provides an opt-out;
• Allows human review of data for training or quality control;
• Gives the provider rights to retain or reuse customer content after termination; or
• Prohibits disclosure of customer data to model providers or other subprocessors.

Recommended Follow-Up

Obtain either:

1. The full redline showing the deleted text; or

2. The prior and revised versions of the agreement.

Particular attention should be given to provisions titled “Data,” “Customer Content,” “Privacy,” “Confidentiality,” “Service Improvement,” “Artificial Intelligence,” “Machine Learning,” “License,” “Training,” “Aggregated Data,” and “Subprocessors.”

Until the deleted language is reviewed, the principal risk is that important customer protections—including restrictions on AI training and secondary use of data—may have been removed without being identifiable from the supplied diff.

Contract Diff Analysis

Key Limitation

The provided diff contains only the statement:

> “Added approximately 4177 words to the document”

It does not include the actual added, deleted, or replaced contractual language. As a result, the legal effect of the changes cannot be reliably assessed.

Customer Data and AI Training

• No specific language is provided addressing:
• Whether customer data may be used to train, fine-tune, or improve AI models;
• Whether inputs, outputs, prompts, personal data, or usage data are retained;
• Whether data may be shared with affiliates, vendors, or model providers;
• Whether data is used for generalized model training or only to provide the services;
• Whether the customer can opt out of AI training or request deletion;
• Ownership or licensing of customer data and AI-generated outputs; or
• Security, confidentiality, anonymization, or de-identification requirements.

Accordingly, it is impossible to determine whether the new text creates or expands the provider’s rights to use customer data for AI training.

Potential Risks Requiring Review

The added language should be examined for provisions that:

1. Grant broad rights to use customer content, prompts, outputs, or telemetry for “improving,” “developing,” or “training” products or models.

2. Permit use of data in aggregated, anonymized, or de-identified form without defining those terms.

3. Allow disclosure to third-party AI providers or subprocessors.

4. Override confidentiality, data-processing, or privacy restrictions.

5. Permit indefinite retention or use after termination.

6. Make the customer responsible for obtaining consent or legal authority for all data submitted.

7. Give the provider ownership of derivative data, model weights, insights, or outputs.

8. Allow policy changes affecting data use without affirmative customer consent.

9. Limit the provider’s liability for unauthorized use, disclosure, or model-training activities.

Conclusion

No substantive legal analysis is possible from the supplied material alone. The complete redline—or at least the approximately 4,177 added words and any corresponding deletions or replacements—is required to identify the important changes and assess AI-training, privacy, confidentiality, ownership, and liability risks.

Summary

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

> “Added approximately 3518 words to the document”

Accordingly, it is not possible to identify:

• The specific legal changes;
• New or expanded customer obligations;
• Changes to liability, indemnity, confidentiality, security, or termination rights;
• Whether the provider receives broader rights to use customer data; or
• Any provisions concerning the use of customer data to train, fine-tune, evaluate, or improve AI models.

AI Training and Data-Use Assessment

No conclusion can be reached from the information provided about whether the contract:

• Permits customer data to be used for AI model training;
• Restricts training to aggregated, anonymized, or de-identified data;
• Allows use of prompts, outputs, files, metadata, or usage data for model improvement;
• Gives the customer an opt-out or requires affirmative consent;
• Allows human review of customer data for training or quality assurance;
• Applies different rules to personal information, confidential information, or regulated data; or
• Requires deletion or exclusion of customer data from future training datasets.

The statement that approximately 3,518 words were added indicates a potentially substantial expansion, but it does not reveal whether those additions are favorable, unfavorable, or unrelated to AI.

Information Needed

Please provide the complete redlined text using the stated notation—for example:

• Additions: {new language}
• Deletions: [deleted language]
• Replacements: [old language]{new language}

Once the actual language is provided, the analysis can identify the material changes, explain their practical effect, and specifically flag any rights to use customer data for AI training or model improvement.

Summary

Key Limitation

The supplied diff states only that approximately 4,177 words were removed. It does not identify which clauses were deleted or provide the original and revised text. Accordingly, it is not possible to determine the specific legal or commercial impact of the changes.

Potential Significance of the Deletions

The removal of a substantial amount of text could affect important provisions, including:

• Customer data ownership and permitted uses
• Confidentiality and data-security obligations
• Privacy-law compliance
• Data retention and deletion
• Service levels, warranties, and disclaimers
• Indemnification and liability limits
• Audit, suspension, and termination rights
• Subcontractor or third-party access
• Intellectual-property ownership
• Governing law and dispute resolution

The deleted material may have narrowed or removed customer protections, but the direction of the change cannot be confirmed from the information provided.

AI Training and Customer Data

There is not enough information to determine whether the agreement’s treatment of customer data for AI training has changed. The deletion may have removed:

• A prohibition on using customer data to train, fine-tune, or improve AI models
• A customer consent or opt-out mechanism
• Restrictions on using prompts, inputs, outputs, or usage data
• Commitments to de-identify or aggregate data before model training
• Limits on retaining data for model-development purposes
• Ownership or licensing terms covering AI-generated outputs
• Disclosure of whether third-party model providers receive customer data
• Security, confidentiality, or deletion requirements applicable to training data

If any such provisions were deleted, the customer could face increased uncertainty or risk that its data, prompts, content, or derived information may be used to develop or improve AI systems.

Recommended Next Step

Provide the actual redline or the text of the deleted provisions, together with the corresponding prior and revised versions. Particular attention should be given to searches for terms such as “train,” “training,” “improve,” “develop,” “machine learning,” “artificial intelligence,” “model,” “inputs,” “prompts,” “customer data,” “usage data,” “de-identify,” “aggregate,” and “retain.”

Summary

The provided diff states only:

> “Added approximately 4172 words to the document”

It does not include the actual added contract language or identify any deletions or replacements.

Substantive Legal Analysis

Because the text of the additions is missing, it is not possible to determine:

• Whether the customer’s data may be used to train, fine-tune, test, or improve AI models;
• Whether such use requires customer consent or is automatic;
• Whether data is anonymized, aggregated, de-identified, or retained in identifiable form;
• Whether customer prompts, outputs, personal information, confidential information, or usage data are included;
• Whether the provider may share data with affiliates, vendors, or model providers;
• Whether the customer can opt out of AI training or revoke consent;
• How long data is retained and whether it is deleted upon termination;
• Whether the provider claims ownership or broad usage rights over customer data or outputs;
• Whether security, confidentiality, regulatory compliance, or indemnity obligations have changed; or
• Whether the new language creates conflicts with existing data-processing, privacy, or confidentiality provisions.

AI-Training Risk Assessment

No conclusion can be reached regarding AI-model training from the information provided. The fact that approximately 4,172 words were added does not, by itself, establish whether customer data may be used for AI training or other model-development purposes.

Information Needed

Please provide the actual redline text, including:

1. All added language shown in {braces};

2. All deleted language shown in [brackets]; and

3. Any replacement language shown in []{braces}.

The full text is necessary to identify new permissions, limitations, consent requirements, opt-out rights, retention periods, and other legal risks.

Summary

The diff states that approximately 3,518 words were removed, but it does not identify which provisions were deleted or provide replacement language. As a result, the specific legal and commercial impact cannot be determined from the information provided.

Key Risks from the Deletions

• Loss of contractual protections: The deleted text may have included limitations of liability, warranties, confidentiality obligations, security commitments, service levels, indemnities, audit rights, or termination rights.
• Changed allocation of risk: Removing provisions may shift risks to the customer, including data-loss, privacy, cybersecurity, intellectual-property, regulatory, or service-disruption risks.
• Ambiguity and interpretation risk: If definitions, exceptions, or procedural requirements were removed, the remaining agreement may be unclear or internally inconsistent.
• Reduced customer remedies: Deleted service credits, refund rights, suspension rights, breach-notice obligations, or dispute mechanisms could materially weaken the customer’s position.
• Regulatory exposure: Deletions concerning data protection, cross-border transfers, retention, subprocessors, or security controls could increase compliance risk.

Customer Data and AI Training

The provided diff contains no text showing whether customer data may be used to train, fine-tune, evaluate, or improve AI models. Therefore, it is not possible to confirm whether the amendment:

• authorizes or prohibits training on customer data;
• expands permitted uses of prompts, inputs, outputs, or usage metadata;
• allows human review or sharing with model providers;
• permits retention of data for model development;
• distinguishes between aggregated, de-identified, and identifiable data;
• grants the provider rights to use customer content after termination; or
• imposes opt-out, deletion, confidentiality, or security protections.

If any AI-related restriction or consent requirement was among the deleted provisions, the deletion could significantly expand the provider’s ability to use customer data. Conversely, deletion of a broad data-use authorization could reduce those rights.

Recommended Review

Obtain the full redline or the deleted text before accepting the amendment. In particular, compare provisions addressing:

1. Customer data ownership and permitted use;

2. AI/model training and improvement;

3. Confidentiality and de-identification;

4. Retention and deletion;

5. Security and subprocessors;

6. Intellectual-property rights; and

7. Liability and indemnification.

No definitive conclusion about AI-training rights or other legal changes can be reached from the deletion summary alone.

Summary

The diff only states that approximately 26 words were removed from the document. The actual deleted language is not provided.

Legal and Commercial Impact

• It is not possible to determine which rights, obligations, limitations, or protections were removed.
• The significance of the change depends entirely on the wording of the deleted text and its location in the agreement.
• Potentially affected areas could include:
• Data ownership and permitted uses
• Confidentiality obligations
• Liability, indemnification, or warranties
• Termination rights
• Security and privacy requirements
• Intellectual-property rights
• Service levels or payment terms

AI Training and Customer Data

• The supplied diff does not identify whether customer data may be used to train, fine-tune, evaluate, or improve artificial-intelligence models.
• It is therefore not possible to determine whether the revision:
• Removes a prohibition on AI training;
• Expands the provider’s rights to use customer data;
• Removes customer consent or opt-out rights;
• Changes whether de-identified, aggregated, or pseudonymized data may be used; or
• Affects deletion, retention, or segregation of customer data used for model development.

Risk Assessment

Assessment: Indeterminate. The omission of the actual 26 deleted words prevents a meaningful legal analysis. The deleted text should be supplied, preferably with the surrounding section and any defined terms, before accepting the change.

No