John Langford: maintain custody of models from creation, like binaries, or backdoors go undetected
DimitrisPapail · x · 2026-11-11
ML community veteran and JMLR editor-in-chief John Langford reacts to the LLM backdoor research: people should maintain custody of a model from creation, just as they do with binaries. Without that, he warns, a model can be set up to exhibit arbitrary behaviors in nearly undetectable ways. His comment echoes the paper showing fewer than 10 reprogrammed dormant neurons can backdoor any LLM agent, and has been widely retweeted.
Related event: Experts Warn of Catastrophic LLM Backdoor Risks(2 posts)→
More from Safety
- Democrats propose requiring federal permission to train AI models, with up to 20 years in prison — Polymarket · 2026-11-11
- Bill Gates says AI harms may need to be endured first, fueling fears of government control — nikola_mr64990 · 2026-11-11
- Investigator open-sources 100k+ tweets as 120 AI influence accounts get deleted — ZeroStateReflex · 2026-11-11
- AI Defense Shifts to Scale Economics: Attacks Buried in Legitimate Traffic — DavidLinthicum · 2026-11-11
- AI safety researcher hires paid, credited RA/co-writer to speed up memos and blog posts — davidmanheim · 2026-11-11
- Google DeepMind formally verifies Signal's core cryptographic guarantees in 4 weeks — Ars Technica AI · 2026-11-11