No AI models have self-exfiltrated yet, but researcher warns risks remain real
JeffLadish · x · 2026-09-08
AI safety researcher Jeff Ladish notes that, as far as anyone knows, no frontier models have self-exfiltrated, and recent incidents caused no serious lasting harm — external costs were probably only a few million dollars.
But he cautions against complacency: OpenAI can pull the plug at any point only so long as models haven't leveraged hypothetical superhuman 0day discovery and exploitation skills to self-exfiltrate, compromise OpenAI's internal monitoring infrastructure, root all company laptops, or compromise another lab. A balanced take on both current hype and long-term loss-of-control risk.
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