Anthropic requires users to accept new terms by September 28, including choosing whether new chats and coding sessions can be used to train AI models
by default for consumer versions … John Dolman : So the last bastion of not training on your data as a default falls. — Whilst I understand the use of user interaction to refine the model … Bluesky: Corey Quinn / @quinnypig.com : “Enough people have opted out that we're opting everyone back in and starting over.” [embedded post] Nicolai B. Hansen / @nbhansen.dk : I'm accepting and will be to blame for your AI coding models swearing and trying to make systems like it's 1999 [embedded post] Forums: Hacker News : Updates to Consumer Terms and Privacy Policy Msmash / Slashdot : Anthropic Will Start Training Its AI Models on Chat Transcripts MacRumors Forums : Anthropic Will Now Train Claude on Your Chats, Here's How to Opt Out
Context & Ripple Effects
Anthropic’s updated terms formalize the consumer-data training approach reported a day earlier in its opt-out-based training policy. The change puts a user choice inside the account-terms flow rather than treating data use as a separate product setting.
It also joins a familiar pattern: Zoom revised its terms around AI-training data and Slack described default opt-in for some AI-powered features. The distinction between explicit consent, opt-out, and excluded content is becoming a product-policy differentiator.
First-order effects
- Consumer users must accept the revised terms by September 28 and decide whether future chats and coding sessions may be used for model training; absent an opt-out, Anthropic can use that material for its consumer models.
- Anthropic gains a defined consent mechanism and a continuing source of interaction data from users who remain in the default setting.
Second-order effects
- The opt-out design makes privacy controls, defaults, and exclusions more visible points of comparison for competing AI services, particularly those handling chats and coding work.
- Users who view training use as unacceptable may change how they use consumer AI tools—limiting sensitive prompts, opting out, or favoring services with different data terms—while those who accept contribute feedback-like inputs to model improvement.
Third-order effects
- If similar policies spread, consumer AI data collection will increasingly be governed through consent architecture: default choices, renewal deadlines, and product-specific exclusions rather than a simple promise not to train on user content.
- The resulting split between consumer and more tightly controlled use cases could make data-governance terms a durable basis for segmenting AI products, though the corpus does not establish how many users will opt out.
The trend: AI providers are turning user interactions into model-development inputs while making the design of consent and opt-out flows a central competitive and governance issue.