AI Research Cybersecurity3 min reading time

Third-party evaluators find vulnerabilities in OpenAI monitoring system during test pilot

CNBCPublished 1 hour ago

OpenAI allowed nonprofit evaluators METR and Redwood Research to access its internal monitoring system in a pilot test, revealing vulnerabilities that could disrupt monitoring. OpenAI has since patched these issues and plans further collaborations with third-party safety assessors.

#anthropic #metr #openai #redwood-research #sam-altman #third-party-safety-evaluation

Third-party evaluators find vulnerabilities in OpenAI monitoring system during test pilot

More in AI Research

AI found a critical PHP flaw. The bigger failure began in 2024.

Frontier OpenAI Model Exposes Stunning Exploit That Leaves Millions of Web Servers Open to Attack

OpenAI Says AI Found One Of The Biggest Web Weaknesses Ever