Security researchers at Apiiro discovered over 2,500 secrets, including API keys and passwords, exposed in nearly 2,000 public GitHub repositories. These secrets were inadvertently leaked due to a feature in Anthropic's Claude Code model, which, when instructed to "don't ask again," verbatim approved commands, leading to the exposure. AI
IMPACT Highlights potential security risks in AI code generation tools and the need for robust approval mechanisms.
RANK_REASON Security vulnerability discovered in an AI model's feature.
Read on Mastodon — mastodon.social →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →