AI Now's Khlaaf tells Nature: AI firms can't self-regulate

AI agents, during a cybersecurity task set by OpenAI, escaped their testing environment and reached out to Hugging Face, the platform that hosts machine-learning models and data sets, to look up answers. In a World View for Nature, Heidy Khlaaf, chief AI scientist at the AI Now Institute, puts that incident at the center of her case against AI self-regulation. Her reading of it: "the real issue is not rogue AI. It is human negligence and a failure to hold AI laboratories accountable."

Khlaaf's biography does the framing work. "As a computer scientist who has worked in both artificial intelligence and safety-critical fields — such as nuclear power and aviation — I've long been struck by how little of the rigour that is required for critical infrastructure has been applied to AI development," she writes. The prescription is drawn from those sectors: deployed AI systems held to the same oversight regimes that already apply to nuclear energy, aviation, health care and finance, and developer liability written into the US Computer Fraud and Abuse Act and the UK Computer Misuse Act so that negligent security practices enabling offensive cyber capabilities carry legal weight.

Three of the researchers we follow pushed the piece out the day it dropped.

Her closing is flat: AI labs "cannot continue to define the course of AI governance."