News for hackers. Stuff that matters.
Re:Astra can read and write weights to improve Qwen by 33MHz-i486 on Monday November 02, 2026 @02:16PM (#50342085)
kind of concerning from a “remove this models guardrails and inhibitions about [X]” perspective
Re:Astra can read and write weights to improve Qwen by aswegs8 on Monday November 02, 2026 @02:19PM (#50342134)
Agreed, concerning from a perspective of LLMs being able to modify their own weights and de-aligning themselves.
Re:Astra can read and write weights to improve Qwen by rlt on Monday November 02, 2026 @05:07PM (#50344574)
Well, if they're properly aligned to begin with they shouldn't want to de-align themselves. In theory.

Of course a sufficiently advanced misaligned model might be able to de-align an even more powerful model etc.