Meta AI and CMU Researchers Cut AI Jailbreak Rates by 22% With Pre-Deployment Training Data Filter
· techtimes.com · popularity 17/100 · 0 similar stories today · AI & Data Science (Artificial Intelligence)
Read the original at techtimes.com →
Summary
LLM fine-tuning safety research from Meta AI and CMU pinpoints the harmless-looking training examples that erode AI model safety before deployment. Researchers from Meta AI, Carnegie Mellon University, and University of Washington built SafeTulu, a filtered dataset that cuts jailbreak attack success rates by up to 22% without hurting mathematical reasoning scores.
Found in: Tech Times. Rating basis: single outlet; outlet tier 2 (unlisted); 1 feeds; feed position 14.