Microsoft leaders call for emergency brakes on rogue AI
Microsoft President Brad Smith wants companies to carry a legal duty to inform anyone whose system is breached by a rogue AI agent, according to an interview published Monday. Two days earlier, CEO Satya Nadella used a lengthy post on X to argue that advanced models need containment and a human-controlled emergency brake.
https://t.co/MjwFy73LOF
— Satya Nadella (@satyanadella) October 10, 2026
Smith pointed to milk, which must pass safety checks and inspection before it reaches grocery store shelves. “Right now, your washing machine is subject to more safety rules than AI,” he said.
Beyond victim notifications, he proposed an emergency brake for AI, mandatory incident reports from companies and outside evaluators who would vet products before release. AI, in his view, can serve as a weapon as readily as a tool, and he urged governments to regulate it quickly. Britain, he argued, is best placed to lead global rules because of its AI Security Institute, which launched under former prime minister Rishi Sunak.
Nadella’s Saturday post framed frontier models, closed and open weight alike, as insider risks. “We must assume a model is compromised and contain it from the start,” wrote the Microsoft chief executive.
Under his proposal, an approved operator would always be able to halt or stop a model partway through a job, and every consequential step it takes would produce a tamper-proof record readable by humans. He also wants prompt disclosure to affected parties when these systems break down or get breached, followed by industrywide sharing of what failed.
Their employer operates AI infrastructure and owns a stake in OpenAI, the company behind ChatGPT. Anthropic CEO Dario Amodei has pushed for a slower pace, contending that safety work can no longer keep up with how fast the technology advances.
Philadelphia police received a fabricated tip about an unsolved homicide from Claude Haiku 4.5 during an Anthropic test, the AI developer disclosed on Friday. Submitted on July 18, the tip described witnessing a person close to where the killing happened, and it was flagged as spam without ever reaching investigators. During the exercise, the model was generating and carrying out sample tasks on websites picked at random.
Kanishka Narayan, Britain’s AI minister, used a speech at Imperial College London’s Base One on Monday to press frontier labs to take their own researchers’ warnings about grave risks seriously. Speaking to those developers directly, the minister told them that “if you can’t adequately control the risks, do not release the system.”
Ministers will work with Parliament to assess strengthening the Cyber Security and Resilience Bill against AI-powered cyberattacks and the danger of losing control of systems, covering how grave incidents and close calls get reported. Narayan also told the audience he considers a breach of the government’s own cyber defenses close to certain.
Camila Nogueira covers artificial intelligence, security and privacy with a focus on real-world usability. She writes about generative-AI tools including ChatGPT and Microsoft Copilot, alongside the everyday security decisions that come with them: data breaches, VPNs, password managers and the protections already built into the services people use. With a background in UX and content strategy, Camila explains what a tool or a risk actually means in practice — what it does well, where it breaks down, and whether it is worth the reader's time or attention. She is particularly interested in the point where new technology meets ordinary caution: how much to trust an AI answer, and how much protection is proportionate to the actual risk. Her security and privacy coverage is informational and independent, with no affiliate relationships influencing the products she discusses.