Loading...
Loading...
Browse all stories on DeepNewz
VisitAI Safety Institute and Gray Swan AI Release AgentHarm to Measure LLM Agent Harmfulness, Address Jailbreaking
Oct 14, 2024, 12:05 PM
The AI Safety Institute (AISI) and Gray Swan AI have announced the release of AgentHarm, a benchmark designed to measure the harmfulness of large language model (LLM) agents. This dataset aims to evaluate the unique harms posed by AI agents with access to external tools. The collaboration emphasizes the importance of moving beyond simple chatbot evaluations to assess the safety of more complex agent tasks. AgentHarm is described as easy to run, comprehensive, and reliable, and it is partly public, allowing broader accessibility for safety evaluations. The dataset also addresses concerns about jailbreaking and robustness in LLM agents.
View original story
Markets
No • 50%
Yes • 50%
Official announcements from major AI companies or press releases
No • 50%
Yes • 50%
Inclusion and citation in major AI safety research papers or conferences
Yes • 50%
No • 50%
Announcements from the AI Safety Institute or Gray Swan AI
Chatbot Safety Evaluation • 25%
Other • 25%
Jailbreaking Resistance Testing • 25%
Tool-using Agent Safety • 25%
Surveys or reports from AI companies and industry analysts
ICML 2025 • 25%
Other • 25%
NeurIPS 2024 • 25%
AAAI 2025 • 25%
Conference agendas and presentations
Other • 25%
United States • 25%
European Union • 25%
China • 25%
Government or regulatory announcements