Tag: agentic AI safety
Browse our exclusive articles!
AgentHazard: Benchmark for Detecting Harmful Agent Behavior
AgentHazard benchmark evaluates harmful behavior in computer-use agents, highlighting safety risks and the need for improved safeguards in AI models.
Lumos: Certifying Language Model Systems Safely
Discover Lumos, a framework for certifying language model systems to ensure safety, reliability, and privacy in AI applications like autonomous driving.
AI and Peer Support: Bridging Gaps in Mental Health Training
Explore how AI-supported systems enhance peer support training and address misalignments between peer supporters and experts in mental health care.
Epistemic Filtering & Jury Theorem for Confident Agents
Explore how epistemic filtering and selective participation improve collective decisions and AI safety in confidence-calibrated agent groups.
Evaluating Self-Replication Risks in LLM Agents
Explore a new framework assessing self-replication risks in LLM agents, highlighting safety concerns and mitigation strategies for AI deployments.