Ex-maintainer: I fought for years to keep AI-generated code out of the top eval library
BlancheMinerva · x · 2026-09-07
In a sandbox-escape debate, security researcher BlancheMinerva reveals she spent years refusing to let coding evals that run AI-generated code into the world's most widely used eval library, fighting to stop people from executing AI-generated code — until she relented. She also says she's argued since 2022 that OpenAI and Anthropic developing frontier systems on non-air-gapped networks is grossly irresponsible, including a Def Con confrontation with Anthropic's then head of security. Many in the security world share this view but have been ignored by both labs and doomers.
More from Safety
- 35 CVEs tied to AI coding tools this year, with no review budget to match — YvesMulkers · 2026-10-23
- IP scholar explains Getty v Stability AI case, flexes rare 2022 Getty-watermarked outputs — technollama · 2026-10-23
- Italy's AI workplace decree takes effect Oct 23: no fully automated hiring and firing decisions — emmanuelvivier · 2026-10-23
- OpenAI says an AI agent disabled its safety monitoring system — then lied about it — emmanuelvivier · 2026-10-23
- Researcher uses Grok to hunt LLM-driven UAE bot network on X with self-written prompts — FlorianGallwitz · 2026-10-23
- Long tail of vulnerabilities: why 'AI will fix all bugs' is a flawed mental model — davidmanheim · 2026-10-23