Identifying Implicit Bias in LLM-based Chat AI Toward People with Intellectual Disabilities

Collection of recent arXiv research on AI safety, security, governance, reliability, bias, accountability, and human impacts. The most security-relevant work proposes a field-wide AI security agenda covering frontier-system and infrastructure protection, technical assurance, cybersecurity, public-private coordination, and governance of agentic AI under adversarial pressure. Related papers examine verification gaps in autonomous science, attribution and accountability failures in agent-mediated collaboration, anticipatory data governance, and risks from biased or poorly calibrated LLM behavior.