Out of Tune: Fine-Tuning Foundation Models Leads to Unpredictable Safety Drift
Recent research we conducted in partnership with researchers at MIT's Computer Science & Artificial Intelligence Laboratory underscores that even ...
Read the original at Center for Democracy & Technology (CDT) ↗
- Topics
- AI Safety & Existential RiskJobs & the Economy
- Mentions
- MIT
- Tracked via
- “AI Risks”