gg2

Opt-out of data for AI training

Archived — this story has rotated out of today’s deck. It is kept here in full.

The gist

Mistral lets users disable AI training via Admin panel or mobile settings.
Opt-out stops new data use but doesn't erase past training or delete data.

Background

AI chatbots often use user conversations to improve models, raising privacy concerns. Mistral provides opt-out controls for its services, while other platforms like Character.AI offer similar settings but with limitations. Legal and regulatory debates continue over consent and copyright in AI training.

How it unfolded

  1. recentlyMistral publishes help article explaining how users can opt out of data training via Admin panel or mobile app settings.
  2. Aug 20, 2026Twitch streamer Warren Pandiscia files class action against Twitch and Amazon, alleging AI training on streams without consent.
  3. Aug 27, 2026CNET reports Stanford study showing six companies use chat data for training, with varying opt-out options.
  4. Sep 2, 2026WIRED reports Trump administration sides with OpenAI in New York Times copyright lawsuit over AI training.

Who’s saying what

Official
Mistral states users retain full control and can opt out of training at any time via provided settings.
Caution
Character.AI's opt-out does not delete existing data or stop use for search, recommendations, and safety classifiers.
Public
Twitch product chief Mike Minton admitted training toggle was opt-out because 'nobody would opt in' if given choice.
Analysts
European Commission's proposed opt-out standards could fragment web and increase compliance costs, harming AI innovation.

Still unverified

Claims that Amazon trained on Twitch streams without consent are allegations in a lawsuit, not yet proven.

Sources

See today’s stories in the app gg2 — free on the App Store