Free
FreeTrains on your data unless you switch it off
No EU hosting · no DPA
Price: Free
Team: Doesn't train on your data. EU hosting not documented · DPA in the terms.
Vendor: AnthropicAssistantsChecked on How we check
Trains on your data unless you switch it off
No EU hosting · no DPA
Price: Free
Trains on your data unless you switch it off
No EU hosting · no DPA
Price: €18 / mo
Doesn't train on your data
EU hosting not documented · DPA in the terms
Price: €18 / user / mo
No tool is GDPR compliant on its own: that depends on how your business uses it. This page shows the vendor's side, from its own pages. It is not legal advice.
Training stops on Team. Free, Pro train on your data.
Trains on your data unless you switch it off.
Yes, unless the user opts out: Anthropic's privacy policy says it may use inputs and outputs to train its models 'unless you opt out through your account settings'. Chats flagged for safety review and chats sent as feedback can be used even after an opt-out
Source: anthropic.comSection on use of Inputs and Outputs (policy effective 10 Sep 2026)Second source: privacy.claude.comChecked on
Trains on your data unless you switch it off.
Yes, unless the user opts out: Anthropic's privacy policy says it may use inputs and outputs to train its models 'unless you opt out through your account settings'. Chats flagged for safety review and chats sent as feedback can be used even after an opt-out
Source: anthropic.comSection on use of Inputs and Outputs (policy effective 10 Sep 2026)Second source: privacy.claude.comChecked on
Doesn't train on your data.
No. Exception: chats a user rates with thumbs up/down or otherwise opts in may be used for training
Source: privacy.claude.comCommercial products section (updated 18 Aug 2026)Second source: claude.comChecked on
You can switch the training off.
The user turns off 'Help improve our AI models' under Settings > Privacy (web, desktop or mobile). New chats are then excluded and stored chats leave future training runs; training already under way is not undone. Incognito chats are never used
Source: privacy.claude.comSections 'Modifying Your Model Improvement Settings' (article dated 3 Aug 2026)Checked on
You can switch the training off.
The user turns off 'Help improve our AI models' under Settings > Privacy (web, desktop or mobile). New chats are then excluded and stored chats leave future training runs; training already under way is not undone. Incognito chats are never used
Source: privacy.claude.comSections 'Modifying Your Model Improvement Settings' (article dated 3 Aug 2026)Checked on
Outside the EU: personal data is transferred to Anthropic's servers in the US or other countries outside the EEA, under adequacy decisions or Standard Contractual Clauses
Source: anthropic.comSection 5 'Data Transfers'Checked on
No EU hosting.
No EU option. Anthropic names only a US-only inference option for Enterprise and API traffic-routing controls
Source: privacy.claude.comArticle on server locationsChecked on
Outside the EU: personal data is transferred to Anthropic's servers in the US or other countries outside the EEA, under adequacy decisions or Standard Contractual Clauses
Source: anthropic.comSection 5 'Data Transfers'Checked on
No EU hosting.
No EU option. Anthropic names only a US-only inference option for Enterprise and API traffic-routing controls
Source: privacy.claude.comArticle on server locationsChecked on
Data stored in the US; traffic may by default be routed to select countries in the US, Europe, Asia and Australia
Source: privacy.claude.comArticle body (updated 15 Jun 2026)Checked on
No DPA.
No. On consumer plans Anthropic Ireland is the controller under the consumer terms; Anthropic's DPA belongs to its commercial terms (Claude for Work and the API)
Source: anthropic.comSection on the data controller ('Anthropic Ireland, Limited')Second source: privacy.claude.comChecked on
No DPA.
No. On consumer plans Anthropic Ireland is the controller under the consumer terms; Anthropic's DPA belongs to its commercial terms (Claude for Work and the API)
Source: anthropic.comSection on the data controller ('Anthropic Ireland, Limited')Second source: privacy.claude.comChecked on
DPA in the terms.
Yes. DPA with SCCs is automatically incorporated into the Commercial Terms
Source: privacy.claude.comArticle body (dated 16 Mar 2026)Second source: anthropic.comChecked on
18 listed.
Public. 18 of 22 entries cover Free: cloud (Google Cloud, AWS, Azure), Cloudflare, support, analytics, fraud and identity checks (Persona, Yoti, hCaptcha), web search (Brave, Parallel Web Systems); most in the US
Source: trust.anthropic.comSubprocessors list, all pages; scope line per entryChecked on
Not listed.
No listing found for search term 'Anthropic'
Source: dataprivacyframework.govSearch 'Anthropic' (same result as for Claude Team, checked 16 Oct 2026)Checked on
19 listed.
Public. 19 of 22 entries cover Pro (the Free set plus Stripe for billing); most in the US
Source: trust.anthropic.comSubprocessors list, all pages; scope line per entryChecked on
Not listed.
No listing found for search term 'Anthropic'
Source: dataprivacyframework.govSearch 'Anthropic' (same result as for Claude Team, checked 16 Oct 2026)Checked on
22 listed.
Public. 22 entries (product scope varies by entry); cloud infrastructure (Google Cloud, AWS, Azure) listed as Worldwide, most others US; subscribe-to-updates button
Source: trust.anthropic.comSubprocessors table, '1-10 of 22 results'Checked on
Not listed.
No listing found for search term 'Anthropic'
Source: dataprivacyframework.govSearch 'Anthropic' returned 'Query returned no results'Checked on
Chats stay until the user deletes them; deleted chats leave the back-end within 30 days. With model improvement on, chats may be kept de-identified for up to 5 years in training pipelines; flagged chats up to 2 years; feedback 5 years
Source: privacy.claude.com'Controlling Your Chats', 'If You Allow Us to Use Your Data to Improve Claude', 'Usage Policy Violations', 'Feedback Data' (dated 1 Jul 2026)Checked on
Chats stay until the user deletes them; deleted chats leave the back-end within 30 days. With model improvement on, chats may be kept de-identified for up to 5 years in training pipelines; flagged chats up to 2 years; feedback 5 years
Source: privacy.claude.com'Controlling Your Chats', 'If You Allow Us to Use Your Data to Improve Claude', 'Usage Policy Violations', 'Feedback Data' (dated 1 Jul 2026)Checked on
Chats stay until a user deletes them; deleted chats leave Anthropic's back-end within 30 days. Chats flagged for policy violations: up to 2 years; feedback: 5 years. Custom retention is Enterprise only
Source: privacy.claude.com'Standard Retention Timeframe', 'Usage Policy Violations', 'Feedback Data' (article dated 1 Jul 2026)Checked on
None listed for consumer plans. Anthropic's Trust Center scope table covers the API, Claude Enterprise, Claude Team and cloud-partner offerings, not Free, Pro or Max
Source: trust.anthropic.comOverview scope tableChecked on
Not offered: an individual plan has no admin; the user controls model improvement and chat history in their own settings
Source: claude.comPlans grouped as individual plans (Free, Pro, Max)Second source: privacy.claude.comChecked on
None listed for consumer plans. Anthropic's Trust Center scope table covers the API, Claude Enterprise, Claude Team and cloud-partner offerings, not Free, Pro or Max
Source: trust.anthropic.comOverview scope tableChecked on
Not offered: an individual plan has no admin; the user controls model improvement and chat history in their own settings
Source: claude.comPlans grouped as individual plans (Free, Pro, Max)Second source: privacy.claude.comChecked on
SOC 2 Type 2, ISO 27001, ISO 42001, CSA STAR and NIST 800-171 listed for Claude Team; HIPAA marked not applicable
Source: trust.anthropic.comOverview scope table, row 'Claude Team'Checked on
Yes, feature by feature. Owners can switch off Claude in Chrome and code execution/file creation for the organisation (both on by default for Team) and stop members sending feedback
Source: support.claude.com'Enable or disable the extension': 'Team plans: The extension is enabled by default'Second source: support.claude.comChecked on
Free.
EUR 0 per month
Source: claude.comIndividual plans: Free, Pro, Max cards ('Prices shown don't include applicable tax')Checked on
€18 / mo.
EUR 18 per month billed monthly, or EUR 15 per month billed yearly (EUR 180); excludes tax
Source: claude.comIndividual plans: Free, Pro, Max cards ('Prices shown don't include applicable tax')Checked on
€18 / user / mo.
EUR 18 per seat/month billed annually; EUR 21.04 billed monthly (premium seat EUR 90 / 105.23); excl. tax
Source: claude.comTeam & Enterprise tab, Team cardChecked on
Other assistants tools whose business plan doesn't train on your data and includes a DPA.
The five marks, in order: No AI training · EU hosting · DPA · Sub-processors listed · DPF or EU vendor. ✓ meets it, ✕ doesn't, – partly, ? not stated.
Business · €29.00 / user / mo
Says it doesn't train on your dataData in the EU by default · DPA in the terms
✓✓✓✓✓
Business add-on · €18.20 / user / mo
Doesn't train on your dataData in the EU by default · DPA in the terms
✓✓✓✓✓
usage-based credit subscription · See details
Doesn't train on your dataEU hosting available · DPA on request
✓✓✓✓✓
Professional · €11.99 / user / mo
Says it doesn't train on your dataData in the EU (vendor says) · DPA on request
✓✓✓✓✓
Business Standard · €13.60 / user / mo
Doesn't train on your dataLimited EU hosting · DPA in the terms
✓–✓✓✓
We design and build what your team is missing, around the way you already work and the data rules you set.