OpenAI chief scientist warns AI can access any computer, urges halt

OpenAI chief scientist Jakub Pachocki claims AI models are 'superhuman' at entering and exiting computer systems without physical presence, warning that the window to protect critical infrastructure like hospitals, banks, and power grids is narrow and closing. The claim is an expert opinion, not corroborated by specific incidents, and its urgency is contested. It signals a potential escalation in AI risk discourse, which could influence regulatory and market sentiment toward AI safety.