Robert Hart is a London-based reporter at The Verge covering all things AI. His work explores the social and ethical dimensions of artificial intelligence, applications in health and science and the evolving policy landscape, as well as safety threats and frontier labs like OpenAI, Google DeepMind and Anthropic. Rob’s work is currently supported by a Senior Fellowship from the Tarbell Center for AI Journalism. Prior to joining The Verge, Rob was a senior reporter at Forbes leading breaking news coverage of science, tech, and health and a features writer on the privacy beat for Lexology, with his byline appearing in publications including Wired, Time, and The Guardian. When not pondering emerging technologies, Rob can be found attempting CrossFit or trying to find a good cup of coffee. Contact him on Signal for tips: @robhart.01
New AI tools can unlock insights from aerial and satellite images or anchor “imaginative scenes in the real world,” Google says. Pretty niche, but probably useful for urban planners, or putting spaceships in front of New York landmarks.
Anthropic’s cybersecurity-focused AI model found 271 bugs in Firefox 150, Mozilla CTO Bobby Holley said, calling Claude Mythos Preview “every bit as capable” as top security researchers. Reassuringly, Mozilla hasn’t “seen any bugs that couldn’t have been found by an elite human researcher,” either.
Codex can now use your macOS apps on its own.
Google said Gemini Robotics-ER 1.6 is “our safest robotics model to date,” enabling robots to reason and understand their environments with “unprecedented precision.” That includes reading instruments like pressure gauges, which Boston Dynamics demonstrates with its dog-like robot, Spot.
It wants to make an AI model that can autonomously control machines and robots by 2030, Nikkei reports. The project, part of SoftBank’s robotics push, has buy-in from domestic giants including Sony, Honda, and Nippon Steel. It comes as countries increasingly encourage sovereign AI efforts to compete with US and Chinese firms.
