Researcher warns policymakers: AI-discovered CVEs at Oracle are the bigger issue

soumitrashukla9 · x · 2026-08-21

Stanford researcher Rishi Bommasani says he routinely references an Epoch plot (plus an aggregate version covering CVEs beyond Oracle) in meetings with policymakers, arguing that AI systems discovering security vulnerabilities is a more fundamental present issue than the OpenAI × Hugging Face incident. The cited thread shows critical vulnerabilities found at Oracle by Mythos and similar AI systems rising sharply, as AI's impact on cybersecurity accelerates.

Original post →

More from Safety

Safety channel →