Good morning. It’s Tuesday, July 28. Anthropic has spent two years selling Claude as the safety first alternative to ChatGPT, and now hundreds of private conversations, some containing clinical trial data and API keys, turned up sitting in a Google search.
Today we are covering:
Private Claude conversations, including sensitive personal data and API keys, were exposed in Google and Bing search results
Anthropic CEO Dario Amodei clarifies the company has never called for banning open-weights AI models, addressing swirling controversy over China AI policy
Microsoft unveils its first AI cybersecurity model, claiming an edge over Anthropic and Google
