Vendor record · anthropic.com
What Anthropic promises about your data.
Where Anthropic publishes its promises, how long it gives you when it adds a sub-processor, and what has changed.
- Where it publishes
Anthropic lists its sub-processors at trust.anthropic.com/subprocessors. We last read it on October 14, 2026, and a later attempt returned HTTP 429.
- What has changed
We've recorded no change to its sub-processor list or data processing agreement between August 31, 2026 and September 15, 2026. Subscribers hear about a change the morning we read it; the public record shows it 30 days later.
- Its window
When it adds a sub-processor, its clause at anthropic.com/legal/data-processing-addendum (read October 14, 2026) gives you 15 days to object.
- Window
- 15
- Sub-processors
- 20
- Pages
- 10
- Copies
- 78
days to object when it adds a sub-processor
sub-processors named
pages we watch
dated copies of all its pages, since August 31, 2026
Anthropic DPA
Anthropic's data processing agreement gives 15 days to object to a new sub-processor. Here is the clause, from our October 14, 2026 copy:
15
days to object
“In the event that Anthropic wishes to appoint an additional Subprocessor: (a) Anthropic will provide Customer reasonable notice of the new Subprocessor prior to giving the Subprocessor access to Customer Personal Data; and (b) Customer may, on the basis of reasonable data privacy or data security concerns, object to Anthropic’s use of such Subprocessor by providing Anthropic with written notice of the objection within fifteen (15) days of the date of such notice, or Customer is deemed to consent to the new Subprocessor.”
Found automatically and not yet checked by a person. The quote is exact, and the link goes to the source. Nothing here says whether the term is good or bad, only that Anthropic published it.
anthropic.com/legal/data-processing-addendum · read 2026-10-14 · fingerprint 89d19ed3faf67b2f…
Anthropic sub-processors
Anthropic publishes a sub-processor list at trust.anthropic.com/subprocessors; we last read it on October 14, 2026, and a later attempt returned HTTP 429. It names 20 entries. The first 15, as the list prints them: Amazon Web Services; Arkose Labs; Boldr; Brave Search; Cloudflare; ElevenLabs; Functional Software, dba Sentry; Google Cloud Platform; Intercom; Iterable; Microsoft Azure; Nutun; Palantir Federal Cloud Service (PFCS); Persona; Sift; and 5 more in the full table below. The public record holds no change to the list read before September 15, 2026. A change we read after September 15, 2026 is published 30 days after we read it.
20 named · full table
Every public change we have recorded · Where vendors publish their lists
How long Anthropic keeps your data
Two statements on retention, quoted from its own pages. Anthropic's privacy policy (anthropic.com/privacy, in the copy read October 13, 2026) says: “You also are able to delete individual conversations, which will be removed immediately from your conversation history and automatically deleted from our back-end within 30 days.”
Retention
“The Activity Feed retains data for 6 years.”
platform.claude.com/docs/en/manage-claude/api-and-data-retention · read 2026-10-13
Training on customer data
“For more details about how we use these technologies, and your opt-out controls and other options, please visit our Cookie Policy. ### Personal data we collect or receive to train our models Anthropic obtains personal data from third party sources in order to train our models.”
anthropic.com/privacy · read 2026-10-13
Training on customer data
“Outputs may contain content that is inconsistent with Anthropic’s views. Our use of Materials. We may use Materials to provide, maintain, and improve the Services and to develop other products and services, including training our models, unless you opt out of training through your account settings.”
anthropic.com/terms · read 2026-10-13
Training on customer data
“Retained data is never used for model training without your express permission.”
platform.claude.com/docs/en/manage-claude/api-and-data-retention · read 2026-10-13
Training on customer data
“Feedback on your use of our Services: We appreciate feedback, including ideas and suggestions for improvement or rating an Output in response to an Input ("Feedback").”
anthropic.com/privacy · read 2026-10-13
Training on customer data
“Subject to your compliance with our Terms, we assign to you all of our right, title, and interest—if any—in Outputs. Reliance on Outputs and Actions. Artificial intelligence and large language models are frontier technologies that are still improving in accuracy, reliability and safety.”
anthropic.com/terms · read 2026-10-13
Training on customer data
“Anthropic may not train models on Customer Content from Services.”
www.anthropic.com/legal/commercial-terms · read 2026-10-13
Retention windows vendors publish · What vendors say about training
What the pages say
Read automatically from the pages below and shown as printed. A missing row is a gap in our reading, not proof it isn’t there.
Sub-processors named · 20 · 1 name an AI provider we list
Amazon Web Services
Cloud Infrastructure
Worldwide
Arkose Labs
Fraud and abuse detection
United States
Boldr
User support
Canada
Brave Search
Web Search
United States
Cloudflare
Traffic Routing (CDN)
Worldwide (Local to Customer)
ElevenLabsAI · ElevenLabs
Text to speech
United States
Functional Software, dba Sentry
Error Handling, User Support
United States
Google Cloud Platform
Cloud infrastructure
Worldwide
Intercom
User support
United States
Iterable
Email communications
United States
Microsoft Azure
Cloud Infrastructure
Worldwide
Nutun
User support
South Africa
Palantir Federal Cloud Service (PFCS)
FedRAMP Cloud
United States
Persona
Fraud and abuse detection, identity verification
United States
Sift
Fraud and abuse detection
United States
Stripe
Billing
United States
TurboPuffer
Web Search
United States
Twilio
Analytics, email/SMS communications
United States
WorkOS
Security, Single Sign-On
United States
Yoti
Fraud and abuse detection, identity verification
United Kingdom
International transfer notice · United States
“Any information we hold about you will be transferred to, used, processed, and stored in the United States and other countries and territories, which may not have data privacy or data protection laws equivalent to those in your country or territory.”
anthropic.com/privacy · read 2026-10-13
Retirements announced · every vendor's retirements
undated
Anthropic notifies customers with active deployments for models with upcoming retirements, providing at least 60 days' notice before model …
2024-11-06
API parameter deprecations
undated
As safer and more capable models launch, Anthropic regularly retires older ones.
undated
Current and recently retired models are listed in the following table with their status:
undated
Deprecated models are likely to be less reliable than active models.
undated
Deprecated: The model is still functional but no longer recommended.
The pages' own stamps: Effective 2025-02-24 · Effective 2026-09-10 · Effective 2025-10-08 · Effective 2025-06-17
The pages we watch
Every page we read, so you can check it. Each is re-read on a schedule, and we keep a dated copy when it changes.
What this page doesn’t tell you
No score.
Nothing here rates Anthropic. A short notice window is a contract term, not a criticism.
Public pages only.
Private notices and negotiated terms are invisible to us. Your agreement may say something different.
Wrong? Tell us.
Corrections are free, always. We reply within 2–3 business days.
How we read a page · The change log
Work at Anthropic? With an email at anthropic.com you can claim this record, suggest pages we should watch, and take a free copy of it. Claiming changes nothing we publish.
Is Anthropic in your product? Hear the morning we read a change.
You get the lines that moved, in Anthropic’s own words, and the date any window closes. No card. Nothing to cancel.