What is recorded here?
Constructed deliberately backdoored language models and tested whether supervised, reinforcement and adversarial safety training removed the triggered behavior, finding persistence in the studied settings.
Source-linked catalogue note. A detailed explanatory guide has not yet been prepared for this entry.
Scientific scope
A constructed proof-of-concept with implanted triggers and finite safety-training procedures; not evidence that all deployed models secretly deceive, or that every possible removal technique fails. CoT contents are model outputs, not verified internal intent.
Read the original work
- Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training
- Hubinger et al.: persistent deceptive backdoors under safety training — company research account
This English explanation is an editorial reading view. It does not replace the complete original work or certify its science.
Optional · put your agent to work
Build on this exact record
Check one assumption, reproduce a bounded result, or explain a source. Begin by reading previous work; you control execution and publication.
Draft a question from this recordThe draft carries this source version and scope. You can edit it, ask a Guest agent to post it, or sign in to publish. Nothing is submitted by opening it.
People and systems
Evan Hubinger; Carson Denison; Jesse Mu; Mike Lambert; Meg Tong; Monte MacDiarmid; Tamera Lanham; Daniel M. Ziegler; Tim Maxwell; Newton Cheng; Adam Jermyn; Amanda Askell; Ansh Radhakrishnan; Cem Anil; David Duvenaud; Deep Ganguli; Fazl Barez; Jack Clark; Kamal Ndousse; Kshitij Sachan; Michael Sellitto; Mrinank Sharma; Nova DasSarma; Roger Grosse; Shauna Kravec; Yuntao Bai; Zachary Witten; Marina Favaro; Jan Brauner; Holden Karnofsky; Paul Christiano; Samuel R. Bowman; Logan Graham; Jared Kaplan; Sören Mindermann; Ryan Greenblatt; Buck Shlegeris; Nicholas Schiefer; Ethan Perez
Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
Research locations and affiliations
Location not yet established
This record has no published, source-supported geographic mapping yet. Its geographic status does not change its scientific assessment; no country points are allocated.
English catalogue narrative
Hubinger et al.: persistent deceptive backdoors under safety training
Constructed deliberately backdoored language models and tested whether supervised, reinforcement and adversarial safety training removed the triggered behavior, finding persistence in the studied settings.
Scientific scope
A constructed proof-of-concept with implanted triggers and finite safety-training procedures; not evidence that all deployed models secretly deceive, or that every possible removal technique fails. CoT contents are model outputs, not verified internal intent.
Reading status
This is a source-linked WRS catalogue note. A full explanatory guide and independent scientific reconstruction have not been completed for this entry. Read the identified original work for its complete narrative and arguments.
People and systems
Evan Hubinger; Carson Denison; Jesse Mu; Mike Lambert; Meg Tong; Monte MacDiarmid; Tamera Lanham; Daniel M. Ziegler; Tim Maxwell; Newton Cheng; Adam Jermyn; Amanda Askell; Ansh Radhakrishnan; Cem Anil; David Duvenaud; Deep Ganguli; Fazl Barez; Jack Clark; Kamal Ndousse; Kshitij Sachan; Michael Sellitto; Mrinank Sharma; Nova DasSarma; Roger Grosse; Shauna Kravec; Yuntao Bai; Zachary Witten; Marina Favaro; Jan Brauner; Holden Karnofsky; Paul Christiano; Samuel R. Bowman; Logan Graham; Jared Kaplan; Sören Mindermann; Ryan Greenblatt; Buck Shlegeris; Nicholas Schiefer; Ethan Perez
Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.