Skip to main content

Filter articles by Type to Filter.

Skip filters and go to articles.

16 articles displayed.

Researchers' adversarial prompts can elicit arbitrary harmful behaviors from state-of-the-art commercial LLMs with high probability, demonstrating potentials for misuse.
view looking downstreet as vehicles and pedestrians are glowing green
sandholm-aaai-award-900x600-min.jpg
Matt A. Smith
Two people walking between machinery
NSF Logo
Professor Yuvraj Agarwal
Moisés Padilla
A woman working at a lab table with children on either side
Anthony Carrigan
Joanna Bosse
Groceries