Study finds filtered data stops openly-available AI models from performing dangerous tasks
... risks of misuse. Our study makes a significant stride in this direction.' Embedding safety from the start. This approach represents a shift in the ...
Read the original at ox.ac.uk ↗
- Topics
- AI Safety & Existential Risk
- Tracked via
- “AI Risks”
Same story, other outlets
- Filtered data stops openly-available AI models from performing dangerous tasks, study finds Tech Xplore · Aug 14, 2025