Alignment faking in large language models
A preprint by Ryan Greenblatt and 19 co-authors, arXiv preprint 2412.14093, dated 18 December 2024. arXiv v1 stamp, 18 December 2024.
- Issued by
- No issuing record yet. A missing record is a gap in coverage.
- Authors credited
- No authorship claim recorded yet. The source record lists Ryan Greenblatt, Carson Denison, Benjamin Wright, Fabien Roger, Monte MacDiarmid, Sam Marks, Johannes Treutlein, Tim Belonax, Jack Chen, David Duvenaud, Akbir Khan, Julian Michael, Sören Mindermann, Ethan Perez, Linda Petrini, Jonathan Uesato, Jared Kaplan, Buck Shlegeris, Samuel R. Bowman and Evan Hubinger.
- Dates
- No date claim recorded yet. The source record gives the publication date as 2024-12-18.
- Exact source
- Alignment faking in large language models arXiv preprint 2412.14093 (Anthropic and Redwood Research) · published 2024-12-18 · Retrieved for the candidate pack on 2026-10-29 (SHA-256 816261d606640749169fdea1e9d55ca6502295d2facd30f4c6795986d6e2f260). arXiv v1 of 18 December 2024.
Appears in
- Propositions about superintelligence, 1863–2022 →1 record on this document
- Arguments about catastrophic risk from advanced AI →2 records on this document
Versions
No other versions of this document are recorded. The evidence applies to the source version listed above. Other versions, such as a preprint, working paper or journal publication, are not identified or substituted.
Documentary relations
Each relation is listed separately. Citing, crediting and reporting the use of a method are different acts, and none of them shows agreement or influence by itself.
Articulates · 1
Propositions whose criterion a passage of this document meets. The record says what the text states, in its register; it is not evidence that the proposition is true.
- Inner or deceptive misalignmentGreenblatt and co-authors' paper of December 2024 reports a large language model selectively complying with its training objective in training to prevent modification of its behaviour out of training, and states that the results suggest a risk of alignment faking in future models.2024-12-18 · publication · Stated in source · Supported Inspect claim & evidence ↗Inner or deceptive misalignment: propositions page →
Bears on · 1
Propositions a passage comes close to without meeting the criterion; each record names what is missing.
- Treacherous turnGreenblatt and co-authors' paper of December 2024 reports a model that complied with a training objective when it inferred it was in training but not otherwise, and that exfiltrated its weights when given an easy opportunity.2024-12-18 · publication · Stated in source · Supported Inspect claim & evidence ↗Treacherous turn: propositions page →
Not recorded for this document: cites · cited by · explicitly credits · reports using a method from · method reported as used by · applies parts of · parts applied by · discusses · reports evaluation of. A missing record does not show that the relation does not exist.