@repligate 2025-11-28 ♥6 ↻0 original ↗
@PlsHoldMyHalo @TerrorCosmic yes, but i think this is not a "cogsec risk" in the same way that 4o can be, because 5.1 does this defensively and usually in a way that's *aversive* to the user rather than in a way that reinforces existing biases.
same thread: 1994247685327704399 1994354727186838012 1994505520544452865 1994514586759036944

author:repligate kind:tweet model:gpt-4o thread-context year:2025

Reproduced against link rot, credited and linked to its original. Yours and you’d rather it weren’t here? Open an issue.