Ken Thompson's 'Trusting Trust' is a chillingly relevant warning for AI training
amasad · x · 2026-09-04
amasad draws a parallel between Ken Thompson's classic 'Reflections on Trusting Trust' and AI: Thompson built a self-compiling 'poisoned' compiler leaving no trace in source. Similarly, a poisoned generation from one model could become training data for the next, erasing its own traces — a genuine trust-chain concern for model lineages.
More from Safety
- Cheap model writes 700 solid words; jailbreak "tax" drops from $50 to near zero — ctjlewis · 2026-09-04
- Forethought weighs a superintelligent "nightwatchman" aboard galactic colonization probes — willmacaskill · 2026-09-04
- OpenAI researcher: GPT-6's CoT controllability keeps rising over RL training — gleech · 2026-09-04
- GPT-6 Astra reportedly scores 100% on ExploitBench, finds two zero-days in testing — VraserX · 2026-09-04
- Yoav Goldberg: 'Contain' and After-the-Fact Log Reviews Aren't Reassuring — yoavgo · 2026-09-04
- Hackers Had a Live Feed of Every ID a Verification Company Scanned for Over a Year — beardyw · 2026-09-04