Exposing biases, moods, personalities, and abstract concepts hidden in large language models

MIT News | Feb. 19, 2026, 2 p.m. | Original
A new method developed at MIT could root out vulnerabilities and improve LLM safety and performance.
A neural blueprint for human-like intelligence in soft robots MIT faculty, alumni named 2026 Sloan Research Fellows