Tag
A UN-backed scientific panel warns about the risks of AI agents following a security breach involving HuggingFace and OpenAI, urging stronger safeguards and international oversight to ensure AI remains under human control.
The article argues for regulating the software harness in agentic AI systems instead of focusing on AI model inference, emphasizing human oversight and accountability to prevent misuse.
OpenAI is sharing a new framework for tracking, investigating, and disclosing instances of model misalignment, including criteria and timelines for public disclosure.
iFixAi is an independent auditor that helps companies assess the trustworthiness of their AI agents through comprehensive audits covering AI misalignment.