Cloud, Alex
Le, Minh https://orcid.org/0009-0003-3214-1051
Chua, James
Betley, Jan https://orcid.org/0009-0008-3518-191X
Sztyber-Betley, Anna https://orcid.org/0000-0002-6464-8194
Mindermann, Sören
Hilton, Jacob https://orcid.org/0009-0002-2002-5929
Marks, Samuel https://orcid.org/0009-0002-3965-4389
Evans, Owain
Article History
Received: 27 July 2025
Accepted: 24 February 2026
First Online: 15 April 2026
Competing interests
: A.C., M.L. and S. Marks are employed by Anthropic, a company that develops LLMs. The remaining authors declare no competing interests.