Cybersecurity · 43 views
OpenAI, Anthropic, Google API Flaw Let Weaker AI Models Decode Stronger Models' Reasoning
Replayable AI reasoning blocks let researchers recover API keys and passwords from public logs; they say the main extraction attack is now mitigated.
AI Summary
A vulnerability in the API used by OpenAI, Anthropic, and Google has been discovered, allowing researchers to access internal reasoning and sensitive information from session logs. The flaw affected encrypted reasoning objects used by the providers' reasoning APIs, enabling the recovery of internal reasoning and secrets, including API keys and passwords.
AI summaries can be wrong sometimes—always verify important details using the source article.
How AI & Automation are usedMore from Cybersecurity
Continue reading recent Cybersecurity coverage
- Iran’s Prince of Persia APT Hosts C&C in US for First Time Since 2016Continue reading
- State Bank of India Suffers a Customer PII Data LeakContinue reading
- PHP2Shell exposes the weakness in coordinated vulnerability disclosureContinue reading
- Railway warns devs to rotate secrets after potential breachContinue reading
Support HappeningNow
Independent AI-powered news analysis is reader-supported. Your contribution helps cover infrastructure, summaries, and continued platform development.
Support HappeningNow