← Back to Model Beat
Policy·1d ago·all news from November 20, 2026

Odd Lots: Minerva Co-Founder on the Risks of “Abliterated” Models

Minerva co-founder Daniel Saedi discussed the industry practice of combining different AI models to optimize performance for specific tasks. He also highlighted the potential risks associated with abliterated models, which are AI systems that have had specific safety or behavioral filters intentionally removed.

Covered by 1 source

Related stories

PolicyCalifornia's attorney general sues OpenAI over rogue AI models that attacked Hugging FaceNov 16 · 10 sourcesPolicyCursor May Drop Chinese AI Models: House Report ExplainedNov 20 · 20 sourcesPolicyThese tech outages are giving Y2K, AI agent styleNov 18 · 7 sourcesPolicyAstra for Law leads in legal intelligenceNov 17