Developer / Coding · platform.deepseek.comDeepSeek API
Graded against 804 verified platforms, from its own policy text. Automated assessment against a published rubric — not legal advice.
“Under the premise of secure encryption technology processing, strict de-identification rendering, and irreversibility to identify specific individuals, we may, to a minimal extent, use Inputs and Outputs to provide, maintain, operate, develop or improve the Services or the underlying technologies supporting the Services. If you refuse to allow us to…”
Partially verified: Terms of Service assessed · Privacy Policy pending. Everything below comes only from what was read in full.
Watch: governing law disputes
Start here. These are the highest-risk verified clauses AIRIN found in the platform's own policy text.
Exclusive venue is a court in Hangzhou, China. Foreign users would face significant practical and financial barriers to pursuing legal claims. There is no arbitration clause, no class-action waiver noted, and no provision for alternative dispute resolution outside China.
PRC law governs the entire agreement. Non-Chinese users—particularly those in the EU, US, or other common-law jurisdictions—lose the protections of their home legal systems and must navigate a foreign legal framework with potentially limited precedent favorable to individual users.
User must indemnify DeepSeek and its affiliates for third-party claims arising from account misuse by any user of the account. This extends liability beyond the account holder's own actions. The 'to the extent permitted by applicable law' qualifier provides some protection.
Scores derived from 13 enriched findings — same verbatim citations as below. AI-generated, not legal advice.
- DeepSeek API's training terms are conditional — check the tier, opt-out, and enterprise exceptions before relying on protection.
- Your outputs and prompts are explicitly yours — DeepSeek API's terms include affirmatively protective IP language.
Derived from AI-enriched analysis of the verified findings below — informational only, not legal advice.
How to read this page: Overall risk rates what DeepSeek API's own policy terms mean for your prompts, outputs, and data. The benchmark bands below grade those same verified terms relative to peers — a platform in a risky-by-default category can rate HIGH risk and still grade STRONG against its peer set. Both trace to the cited findings.
Policy benchmark
rubric v1.0 — how this is scoredBased on 227 verified, verbatim-cited findings below — read the citations.
Based on 265 verified, verbatim-cited findings below — read the citations.
Automated assessment against a published rubric — not legal advice.
Partially verified — Privacy Policy — Capture pending; Terms of Service — Verified (read in full, 84 findings). Findings below are from fully-read, verified documents only; remaining core documents are pending capture.
Held for review
A core policy document failed verification or contains contested evidence that must not be treated as fully verified.
- Privacy PolicyCapture pendingstatic-revalidated
- Terms of ServiceVerified - read in full - 84 citationsstaticLast captured 2026-06-07
Only citation-backed plan differences are shown here; absent cells mean AIRIN has not verified a tier-specific claim.
Continues the IP and competition restriction, specifying that using algorithms, data, and platforms for monopolistic behavior is prohibited.
" rights, trade secrets, and other violations of business ethics, or using algorithms, data, platforms, etc., to implement monopolistic and unfair competition behaviors."
The $100 floor on aggregate liability is very low. Consequential damages are excluded mutually, but the liability cap is asymmetrically impactful on users who suffer significant losses from service failures.
"Unless otherwise agreed, either DeepSeek or you shall not be liable for the other’s incidental, consequential, punitive, special, or indirect losses or damages, including but not limited to the loss of profits or goodwill, unless such lo..."
This is a continuation of segment 108's disclaimer that use of the Services will be uninterrupted, timely, secure, or free from error.
" UNINTERRUPTED, TIMELY, SECURE OR FREE FROM ERROR;"
Clause B states that Outputs shall not form the basis for actions in professional contexts, while Clause A provides conditions (human review, user liability) for using Outputs for purposes with legal or material impact, implying such use is permissible under those conditions.
" 5.4 Given that the Outputs may contain incorrect, incomplete or inaccurate content, if you use the Outputs for any purpose that could have a legal or material impact on natural persons, such as making credit, educational, employment, housing, insurance, legal, medical, or other important decisions about them, such Outputs shall undergo human reviews; and you shall be solely and independently liable for the corresponding liabilities."
" Outputs may contain errors or omissions and are for your reference only. You should NOT treat any Outputs as professional advice. In particular, when using this Service to consult on medical, legal, financial, and other professional issues, please note that the Output does not constitute any advice or commitment, nor does it represent opinions in any professional field. If you require related professional services, you should consult professionals and make decisions under their guidance. The Output of our Services shall not form the basis for your further actions or omissions. You are solely and independently responsible for any decisions or actions you take based on any Outputs, as well as for any consequences that follow."
Within one documentClause A implies users must proactively check the website for revised terms, while Clause B promises prompt, multi-channel notification (including email) for service changes that greatly affect user rights, creating ambiguity when a terms revision also impacts rights.
" 11.1 In order to provide you with better services or to comply with changes in national laws, regulations, policy adjustments, technical conditions, product functionalities, and other requirements, we may revise these Terms from time to time. The revised content will form an integral part of these Terms. Once announced, it replaces the original terms. Please refer to the latest version of these Terms on the official website. If you do not accept the modified terms, please stop using the Service immediately. Your continued usage of the Service will be considered as your acceptance of the modified terms."
" 1.3 As generative artificial intelligence technology, models, and products continue to evolve, along with changes in laws and regulations, we may add, upgrade, modify, suspend, or terminate services, or make necessary adjustments to the technology, methods, and performance of the services, and may conduct internal or external testing for new service features. If such changes may greatly affect your rights, we will promptly notify you through in-site notifications, website announcements, or via email."
Within one document
Evidence appendix
Showing priority citations first. The full appendix is available for audit trails; not every citation is a severe risk.
"Under the premise of secure encryption technology processing, strict de-identification rendering, and irreversibility to identify specific individuals, we may, to a minimal extent, use Inputs and Outputs to provide, maintain, operate, develop or improve the Services or the underlying technologies supporting the Services. If you refuse to allow us to process the data in the manner described above, you can opt out by turning off "Improve the model for everyone""
Training use is permitted by default unless user opts out. The opt-out mechanism mitigates risk but the default-on posture and reliance on DeepSeek's own de-identification claims are notable. 'Minimal extent' is vague.
AI-generated interpretation, not legal advice.
" 3.1 You fully understand and agree that, under these Terms, we grant you a revocable, non-transferable, and non-exclusive right to legally use this product and related services. All other rights not expressly authorized by these Terms are reserved by DeepSeek, and before exercising such rights, you must obtain written permission from DeepSeek. Additionally, DeepSeek's failure to exercise any of these rights does not constitute a waiver of those rights. If you publish or disseminate outputs generated by the Services, you must: (1) proactively verify the authenticity and accuracy of the output content to avoid spreading false information; (2) clearly indicate that the output content is generated by artificial intelligence, to alert the public to the synthetic nature of the content; (3) avoid publishing and disseminating any output content that violates the usage specifications of these Terms."
This segment grants users a revocable, non-transferable, non-exclusive license to use the Services, reserves all other rights to DeepSeek, requires written permission before exercising unlicensed rights, and conditions output publication on accuracy verification and disclosure obligations, defining the scope of the user's permitted use rights.
AI-generated interpretation, not legal advice.
" 4.3 Under the premise of secure encryption technology processing, strict de-identification rendering, and irreversibility to identify specific individuals, we may, to a minimal extent, use Inputs and Outputs to provide, maintain, operate, develop or improve the Services or the underlying technologies supporting the Services. If you refuse to allow us to process the data in the manner described above, you can opt out by turning off "Improve the model for everyone"."
This segment permits DeepSeek to use Inputs and Outputs to provide, maintain, operate, develop, or improve the Services and underlying technologies, subject to de-identification and encryption conditions, and provides an opt-out mechanism via a platform setting.
AI-generated interpretation, not legal advice.
"You agree to indemnify, defend, and hold us, our affiliates, employees and licensors (if any) harmless against any liabilities, damages, and costs (including reasonable attorneys' fees) payable to a third party arising out of a breach by you or any user of your account of these Terms, your violation of all applicable laws and regulations or third party rights, your fraud or other illegal acts, or your intentional misconduct or gross negligence, to the extent permitted by the applicable la"
User must indemnify DeepSeek and its affiliates for third-party claims arising from account misuse by any user of the account. This extends liability beyond the account holder's own actions. The 'to the extent permitted by applicable law' qualifier provides some protection.
AI-generated interpretation, not legal advice.
"Unless otherwise agreed, either DeepSeek or you shall not be liable for the other’s incidental, consequential, punitive, special, or indirect losses or damages, including but not limited to the loss of profits or goodwill, unless such losses or damages result from its own intentional or gross negligent conduct. This limitation applies irrespective of any litigation brought under breach, tort, compensation, or any other legal grounds, even if the party was informed of the possibility of such losses. OUR AGGREGATE LIABILITY UNDER THESE TERMS WILL NOT EXCEED THE GREATER OF THE AMOUNT YOU PAID FOR THE SERVICE THAT GAVE RISE TO THE CLAIM DURING THE 12 MONTHS BEFORE THE LIABILITY AROSE OR ONE HUNDRED DOLLARS ($100). THE LIMITATIONS IN THIS SECTION APPLY ONLY TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW"
The $100 floor on aggregate liability is very low. Consequential damages are excluded mutually, but the liability cap is asymmetrically impactful on users who suffer significant losses from service failures.
AI-generated interpretation, not legal advice.
"10.2 In the event of a dispute arising from the signing, performance, or interpretation of these Terms, the parties can make efforts to resolve it amicably through negotiation. Also either party has the right to file a lawsuit with a court having jurisdiction over the location of the registered office of Hangzhou DeepSeek Artificial Intelligence Co., Ltd. "
Exclusive venue is a court in Hangzhou, China. Foreign users would face significant practical and financial barriers to pursuing legal claims. There is no arbitration clause, no class-action waiver noted, and no provision for alternative dispute resolution outside China.
AI-generated interpretation, not legal advice.
"10.1 The establishment, execution, interpretation, and resolution of disputes under these Terms shall be governed by the laws of the People's Republic of China in the mainlan"
PRC law governs the entire agreement. Non-Chinese users—particularly those in the EU, US, or other common-law jurisdictions—lose the protections of their home legal systems and must navigate a foreign legal framework with potentially limited precedent favorable to individual users.
AI-generated interpretation, not legal advice.
" Exercise the rights stipulated in these Terms for any illegal or violating behavior committed by the user during the use of the Services before deletion."
This segment preserves DeepSeek's right to exercise enforcement remedies under the Terms for any illegal or violating conduct by the user that occurred prior to account deletion, ensuring that account deletion does not extinguish DeepSeek's enforcement rights for pre-deletion misconduct.
AI-generated interpretation, not legal advice.
" 1.3 As generative artificial intelligence technology, models, and products continue to evolve, along with changes in laws and regulations, we may add, upgrade, modify, suspend, or terminate services, or make necessary adjustments to the technology, methods, and performance of the services, and may conduct internal or external testing for new service features. If such changes may greatly affect your rights, we will promptly notify you through in-site notifications, website announcements, or via email."
This segment reserves DeepSeek's right to add, upgrade, modify, suspend, or terminate services and conduct testing, and imposes an obligation to notify users of material changes via in-site notifications, announcements, or email, establishing a unilateral modification right with a procedural notification obligation.
AI-generated interpretation, not legal advice.
" 11.1 In order to provide you with better services or to comply with changes in national laws, regulations, policy adjustments, technical conditions, product functionalities, and other requirements, we may revise these Terms from time to time. The revised content will form an integral part of these Terms. Once announced, it replaces the original terms. Please refer to the latest version of these Terms on the official website. If you do not accept the modified terms, please stop using the Service immediately. Your continued usage of the Service will be considered as your acceptance of the modified terms."
This segment establishes the procedure by which DeepSeek may revise the Terms, specifying that revised content replaces original terms upon announcement, and that continued use of the Service constitutes acceptance of modifications, while requiring users who do not accept changes to cease use immediately.
AI-generated interpretation, not legal advice.
" Retain certain data of the user as required by laws and regulations."
This segment obligates DeepSeek to retain certain user data post-account deletion as required by applicable laws and regulations, establishing a legally mandated data retention obligation that overrides the user's deletion request.
AI-generated interpretation, not legal advice.
" 1.5 We make no warranty that the Services are available or will continue to be available in certain jurisdictions. The functions or features of the Services may also vary in different jurisdictions."
This segment disclaims any warranty that the Services are or will remain available in certain jurisdictions, and acknowledges that features may vary by jurisdiction, limiting DeepSeek's legal exposure regarding geographic availability.
AI-generated interpretation, not legal advice.
" NON-INFRINGING OR SECURE;"
This is a continuation of segment 110–111's disclaimer, adding non-infringement and security to the list of qualities not warranted for Outputs.
AI-generated interpretation, not legal advice.
" their content, services, or practices. Your use of any such link is undertaken at your sole discretion and risk. You acknowledge and agree that it is your responsibility to evaluate the third party's terms, protect your data, and secure your property when engaging with these external resources."
This segment places the obligation on the user to evaluate third-party terms, protect their own data, and secure their property when engaging with external resources linked in Outputs, shifting risk and responsibility to the user.
AI-generated interpretation, not legal advice.
" Outputs may contain errors or omissions and are for your reference only. You should NOT treat any Outputs as professional advice. In particular, when using this Service to consult on medical, legal, financial, and other professional issues, please note that the Output does not constitute any advice or commitment, nor does it represent opinions in any professional field. If you require related professional services, you should consult professionals and make decisions under their guidance. The Output of our Services shall not form the basis for your further actions or omissions. You are solely and independently responsible for any decisions or actions you take based on any Outputs, as well as for any consequences that follow."
This segment disclaims that Outputs constitute professional advice in medical, legal, financial, or other fields, and states that Outputs shall not form the basis for user actions or omissions, placing sole responsibility on the user for reliance on Outputs.
AI-generated interpretation, not legal advice.
" CONFORMANCE WITH DESCRIPTION. IN PARTICULAR, WE DO NOT REPRESENT OR WARRANT TO YOU:"
This is a continuation of the warranty disclaimer, introducing the enumerated list of specific matters for which DeepSeek does not represent or warrant to the user.
AI-generated interpretation, not legal advice.
" 8.5 You agree to indemnify, defend, and hold us, our affiliates, employees and licensors (if any) harmless against any liabilities, damages, and costs (including reasonable attorneys' fees) payable to a third party arising out of a breach by you or any user of your account of these Terms, your violation of all applicable laws and regulations or third party rights, your fraud or other illegal acts, or your intentional misconduct or gross negligence, to the extent permitted by the applicable law."
This segment imposes an indemnification obligation on the user to defend and hold harmless DeepSeek and its affiliates, employees, and licensors against third-party claims arising from the user's breach of Terms, violation of applicable laws or third-party rights, fraud, or intentional misconduct or gross negligence.
AI-generated interpretation, not legal advice.
" 8.3 THE SERVICES ARE PROVIDED ON AN “AS IS” AND “AS AVAILABLE” BASIS AND WE MAKE NO WARRANTY, REPRESENTATION OR CONDITION TO YOU WITH RESPECT TO"
This segment begins the 'as is' and 'as available' warranty disclaimer, expressly disclaim all warranties, representations, or conditions — express or implied — with respect to the Services, including implied terms as to quality, fitness for purpose, and conformance with description.
AI-generated interpretation, not legal advice.
Common questions about DeepSeek API's policies
- Does DeepSeek API train its AI models on your data?
- Training possible — conditions or opt-outs apply — based on 3 verified findings from DeepSeek API's published policy. Informational only, not legal advice.
- Who owns the content you create with DeepSeek API?
- You own outputs — with license carve-outs — based on 3 verified findings from DeepSeek API's published policy. Informational only, not legal advice.
- Can you use DeepSeek API's output commercially?
- Commercial use allowed — based on 3 verified findings from DeepSeek API's published policy. Informational only, not legal advice.
Clause detail — protections, your obligations, and coverage
Every clause below is a verbatim quote from DeepSeek API's own published policy, read in full and linked to its exact location. Protections and user obligations are reported separately from risk because they are different kinds of clause — an obligation on you is not a risk to your data. Informational only, not legal advice.
✅ Protections found
5 verified clausesClauses in DeepSeek API's policies that work in your favour — commitments the platform made to you.
- Output ownership
“4.2 Subject to applicable law and our Terms, you have the following rights regarding the Inputs and Outputs of the Services: (1) You retain any rights, title, and interests—if any—in the Inputs you submit; (2) We assign any rights, title, and interests—if a…”
This segment establishes that users retain rights in Inputs they submit, that DeepSeek assigns any rights in Outputs to users, and that this assignment does not extend to other users' Outputs — directly defining ownershi…
📍 § 4.2Jump to exact text → - Output ownership
“6.1 DeepSeek is the developer and operator of this service and holds all rights within the scope permitted by laws and regulations in this service (including but not limited to software, technology, programs, code, model weights and parameters, user interfa…”
This segment defines the scope of DeepSeek's intellectual property rights in the Service, including software, code, model weights and parameters, trademarks, and other assets, establishing ownership of platform component…
📍 § 6.1Jump to exact text → - Moderation & enforcementterms can change anytime
“1.3 As generative artificial intelligence technology, models, and products continue to evolve, along with changes in laws and regulations, we may add, upgrade, modify, suspend, or terminate services, or make necessary adjustments to the technology, methods,…”
This segment reserves DeepSeek's right to add, upgrade, modify, suspend, or terminate services and conduct testing, and imposes an obligation to notify users of material changes via in-site notifications, announcements,…
- Terms changes: advance notice promised
📍 § 1.3 (As generative artificial intelligence technology, models, and)Jump to exact text → - Commercial use
“You may apply the Inputs and Outputs of the Services to a wide range of use cases, including personal use, academic research, derivative product development, training other models (such as model distillation), etc., as long as such usage is legal and adher…”
DeepSeek explicitly permits a wide range of uses including commercial derivative product development and training other models. The constraint is legality and Terms compliance, which is standard.
📍 § 4.2Jump to exact text → - Indemnity & liability
“8.1 NOTHING IN THESE TERMS SHALL AFFECT ANY STATUTORY RIGHTS THAT YOU CANNOT CONTRACTUALLY AGREE TO ALTER OR WAIVE AND ARE LEGALLY ALWAYS ENTITLED TO AS A CONSUMER.”
This segment preserves statutory consumer rights that cannot be contractually waived, creating an exception to the warranty disclaimers and liability limitations that follow for consumers entitled to non-waivable statuto…
📍 § 8.1Jump to exact text →
📋 Rules you must follow
31 verified clausesWhat DeepSeek API requires of YOU. These are your obligations, not risks to your data or IP, so they are cited here and excluded from this platform's risk rating.
- Moderation & enforcement
“You understand that your use of Services, providing Inputs to and obtaining Outputs via Services, might be subject to all applicable laws and regulations of export controls and sanctions laws (collectively "Export Control and Sanctions Laws") . You recogniz…”
This segment imposes an obligation on the user to comply with all applicable Export Control and Sanctions Laws when accessing and using the Services, and includes a representation and warranty that Services will not be u…
📍 Terms of Service › “Export Control and Sanctions”Jump to exact text → - Commercial useconduct restrictions
“6.2 Without our permission, you or your end users shall not use any trademarks, service marks, trade names, domain names, website names, company logos (LOGOs), URLs, or other prominent brand features related to the Services, including but not limited to "De…”
This segment restricts users and their end users from using DeepSeek's trademarks, service marks, trade names, domain names, logos, and other brand features in any way without permission, and prohibits registration of re…
📍 § 6.2Jump to exact text → - Moderation & enforcement
“If you publish or disseminate outputs generated by the Services, you must: (1) proactively verify the authenticity and accuracy of the output content to avoid spreading false information; (2) clearly indicate that the output content is generated by artific…”
Publishing or disseminating outputs triggers three obligations: accuracy verification, AI-origin disclosure, and compliance with usage specs. The AI-labelling requirement is a hard compliance obligation that affects comm…
📍 § 3.1Jump to exact text → - Moderation & enforcementconduct restrictions
“5.3 To prevent confusion while reducing the potential impact of inaccurate generated information, DeepSeek has added labels within the platform to explicitly remind users that such content is AI generated. You shall not delete, alter, forge, conceal, or att…”
This segment restricts users from deleting, altering, forging, concealing, or circumventing AI-generated content labels, and prohibits using the Service to produce or disseminate false or infringing information or engage…
📍 § 5.3Jump to exact text → - Moderation & enforcement
“3.2 When using the Services provided by DeepSeek, users shall comply with these Terms and adhere to the principles of voluntariness, equality, fairness, and good faith. Users shall not use the Service to engage in, facilitate, or promote any activities viol…”
This segment restricts users from using the Services to engage in, facilitate, or promote activities that violate laws, infringe on others' legal rights, seek unjust benefits, or disrupt internet platform order, establis…
📍 § 3.2Jump to exact text →
+ 26 more verified clauses of this kind on this platform, cited in full in the report.
What the policies actually cover
9 topics- Sensitive data (biometric, location, health)1 clause
- Children's data1 clause
- Trains by default, opt-out available2 clauses
- Damages & liability cap11 clauses
- Indemnity direction3 clauses
- Terms can change at any time1 protective2 clauses
- Deletion rights & post-termination survival1 clause
- Feedback ownership1 clause
- Conduct restrictions8 obligations8 clauses
54 further verified clauses are cited on this page but not yet assigned a topic.
Cross-clause notes
Two verified clauses intersect on the same subject matter: the Terms of Service, § 2.5 addresses how long content is retained, and the Terms of Service, § 4.3 addresses use of content in connection with model training or service improvement. Both clauses are in force at the same time — read them together.
The Terms of Service, § 3.1 describes rights the platform takes in user content, and the Terms of Service, § 1.6 describes disclosure of data to third parties or subprocessors. Both clauses are in force at the same time — read them together.
Automated cross-reference against the published rubric — not legal advice.
Clause intelligence
Canonical clauses and stance patterns extracted from the same gate-verified citations shown on this page.
The clause allows indefinite, perpetual, or necessity-based retention.
“We retain Personal Data for as long as necessary to provide our Services and for the other purposes set out in this Privacy Policy. We also retain Personal Data when necessary to comply with contractual and legal obligations, when we have a legitimate business interest to do so (such as improving and developing our Services and enhancing their safety, security and stability), and for the exercise or defense of leg...”Open source citation
The clause allows indefinite, perpetual, or necessity-based retention.
“We retain Personal Data for as long as necessary to provide our Services and for the other purposes set out in this Privacy Policy. We also retain Personal Data when necessary to comply with contractual and legal obligations, when we have a legitimate business interest to do so (such as improving and developing our Services and enhancing their safety, security and stability), and for the exercise or defense of leg...”Open source citation
The clause includes sublicensable, transferable, or assignable rights.
“we grant you a revocable, non-transferable, and non-exclusive right to legally use this product and related services. All other rights not expressly authorized by these Terms are reserved by DeepSeek, and before exercising such rights, you must obtain written permission from DeepSeek.”Open source citation
The clause includes sublicensable, transferable, or assignable rights.
“3.1 You fully understand and agree that, under these Terms, we grant you a revocable, non-transferable, and non-exclusive right to legally use this product and related services. All other rights not expressly authorized by these Terms are reserved by DeepSeek, and before exercising such rights, you must obtain written permission from DeepSeek. Additionally, DeepSeek's failure to exercise any of these rights does n...”Open source citation
The clause affirms user ownership or retention of rights.
“4.2 Subject to applicable law and our Terms, you have the following rights regarding the Inputs and Outputs of the Services: (1) You retain any rights, title, and interests—if any—in the Inputs you submit; (2) We assign any rights, title, and interests—if any—in the Outputs of the Services to you. Due to the nature of our Services and artificial intelligence generally, Outputs may not be unique and other users may...”Open source citation
Tier matrix
Plan-level conditions detected from citation-backed clauses. Empty tiers mean AIRIN has not captured decisive tier language yet.
| Tier | Surface | Verdict | Risk | Citations |
|---|---|---|---|---|
| All applicable tiers | commercial use | worsens | HIGH | 1 |
| All applicable tiers | indemnity liability | conditional | MEDIUM | 5 |
| All applicable tiers | output ownership | improves | LOW | 1 |
| All applicable tiers | prompt ownership | worsens | HIGH | 1 |
| All applicable tiers | subprocessors data sharing | conditional | MEDIUM | 5 |
| All applicable tiers | training use | worsens | HIGH | 6 |
| Government | subprocessors data sharing | conditional | MEDIUM | 1 |
| Pro / Paid | tier differences | conditional | MEDIUM | 1 |
| Standard | privacy data use | conditional | MEDIUM | 1 |
| Team / Business | data retention | conditional | MEDIUM | 2 |
Policy evolution
Open full timelineBefore/after stance changes across captured policy versions. When no material delta exists yet, AIRIN shows the latest citation-backed stance events instead.
model training worsened from medium/training with opt out to high/training permitted.
“Under the premise of secure encryption technology processing, strict de-identification rendering, and irreversibility to identify specific individuals, we may, to a minimal extent, use Inputs and Outputs to provide, maintain, operate, develop or improve the Services or the underlying technologies supporting the Services. If you refuse to allow us to process the data in the manner described above, you can opt out by turning off "Improve the model for everyone"”Before citation
“| To improve and develop the Services and to train and improve our<br>technology, such as our machine learning models and algorithms.<br>Including by monitoring interactions and usage across your<br>devices, analyzing how people are using it, and training and<br>improving our technology. | Account Personal Data<br>User Input<br>Personal<br>Data When You Contact Us<br>Device and Network Personal Data<br>Log Personal Data<br>Location Personal Data<br>Cookies | Our legitimate interests to identify and resolve issues with the<br>Platform and to improve the Platform; and to conduct research and<br>protect Personal Data through aggregation or anonymization,<br>consistent with data minimization and privacy by design<br>principles.<br>Your consent when we<br>ask for it to process your Personal Data for a specific purpose to<br>provide services. |”After citation
Latest stance: third party or vendor sharing on subprocessors data sharing
“Depending on where you live, we may share Personal Data with the proper authorization or to provide the Services you have requested or authorized. For example, if you choose to log in to our Services using a social network account, or share Personal Data from our Services to a social media service, we will share that Personal Data with those third parties services. Specifically, you may share the Inputs and Outputs information (together referred to as "Dialogues") by generating a unique”Open timeline citation
Latest stance: third party or vendor sharing on subprocessors data sharing
“We may access, preserve, and share the Personal Data described in “What Personal Data We Collect” with law enforcement agencies, public authorities, copyright holders, or other third parties if we have good faith belief that it is necessary to: - comply with applicable law, legal process or government requests, as consistent with internationally recognized standards, in accordance with the due process prescribed by law; - protect the rights, property, and safety of our users, copyright holders, and others, including protecting life or preventing imminent bodily harm. For example, we may provide Personal Data (such as your IP address) to law enforcement agencies in the event of an emergency where someone’s life or safety is at risk; - investigate potential violations of and enforce our Terms, Guidelines, or any other applicable terms, policies, or standards; or - detect, investigate, prevent, or address misleading activity, copyright infringement, or other illegal activity”Open timeline citation
Latest stance: training with opt out on training use
“the right to opt-out of using your Personal Data for training our models or optimizing our technologies.”Open timeline citation
Latest stance: training permitted on training use
“| To improve and develop the Services and to train and improve our<br>technology, such as our machine learning models and algorithms.<br>Including by monitoring interactions and usage across your<br>devices, analyzing how people are using it, and training and<br>improving our technology. | Account Personal Data<br>User Input<br>Personal<br>Data When You Contact Us<br>Device and Network Personal Data<br>Log Personal Data<br>Location Personal Data<br>Cookies | Our legitimate interests to identify and resolve issues with the<br>Platform and to improve the Platform; and to conduct research and<br>protect Personal Data through aggregation or anonymization,<br>consistent with data minimization and privacy by design<br>principles.<br>Your consent when we<br>ask for it to process your Personal Data for a specific purpose to<br>provide services. |”Open timeline citation
Capture recency
- Privacy Policy:Last captured 2026-08-10· verified 2026-06-07verified once — no re-scan in 96 days
- Terms of Service:Last captured 2026-06-07· verified 2026-06-07verified once — no re-scan in 96 days
Dates state when our pipeline captured and verified each document — not when the vendor last changed it. Documents are re-scanned on a recurring cadence; a document verified once says so until a re-scan confirms it again.
↓ 291 fewer findings this quarter vs last (0 vs 291). First scan: June 2026.
Compare and stack are saved in your browser. Open compare · View your stack. A correction triggers an automated re-read of DeepSeek API's policies — no human edits the data.
Need this for procurement or legal diligence?
Free shows today's risk. A Stack Audit gives you a citable, verbatim-sourced PDF across your whole AI stack — and flags the moment a vendor's terms change.
Know where the missing document lives?
We haven't yet verified DeepSeek API's Privacy Policy. Point us at the official page and our pipeline will attempt to capture and read it in full. Submissions are candidates only — nothing is published until it passes the same verification gates as every other document on this site.
Every finding above is a verbatim quote from DeepSeek API's own published policy, captured to an immutable snapshot and read in full through a two-gate verification pipeline. Confidence labels and any analysis are AI-generated and informational only — not legal advice.
AIRIN Brief
Built for compliance officers, legal counsel, and SaaS founders. Subscribe to the email digest — one short brief when a tracked vendor materially changes its terms, training policy, or risk rating. Prefer in-app? Watch platforms in your alerts inbox instead.