THE RUNDOWN

Over a thousand employees across the biggest AI labs have written to the US government asking it to do something about runaway AI automation. The letter asks for minimum safety standards and clearer oversight frameworks. It's a rare moment of the industry turning around and asking to be regulated — driven, it seems, by genuine anxiety about where things are heading. You can find deeper context on this at our AI coverage.

Sam Altman's willingness to decelerate marks a real shift in tone from the world's most influential AI CEO. He hasn't named the security incident that changed his thinking, but the fact that he's speaking openly about feeling shaken is worth paying attention to. Whether OpenAI's actual pace slows is another question entirely.

It's earnings season, and Google handed Wall Street a nasty surprise — a revised capital expenditure estimate of up to $205 billion, up sharply from last quarter's already-enormous figure. Investors aren't necessarily against AI investment, but a number that size raises real questions about when it translates into returns. AI is now expensive enough to make the money people genuinely nervous.

The US government has announced an import ban on 'advanced robotic devices' and power inverters manufactured abroad — with China squarely in its sights. The move is framed as a national security strategy to kickstart domestic production of emerging technology, including humanoid robots. It's another escalation in the tech cold war, and it will have ripple effects for anyone building hardware supply chains that touch both markets.

Cybersecurity company Cyera is acquiring Oasis Security in a $1 billion deal focused specifically on protecting AI agents — the autonomous software bots increasingly running sensitive business tasks. It's Cyera's third acquisition this year, which tells you something about how fast the market for AI security is moving. As AI agents get more powerful, keeping them from being exploited becomes a serious business.

This one is genuinely wild. Back in 2019, senior eBay executives orchestrated a harassment campaign against Ina and David Steiner, a Massachusetts couple who ran an e-commerce newsletter and occasionally criticised the company. The campaign included cyberstalking, physical surveillance, and sending disturbing packages to their home. eBay and three former executives have now agreed to pay $55.7 million to settle the case. It's a long-overdue ending to one of the stranger corporate misconduct stories in recent memory.

A fired Tesla manager has filed a lawsuit claiming the company's Full Self-Driving robotaxis were 'rolling hazards' and that safety operators overseeing them were dangerously overstretched. The allegations add fuel to an already heated debate about whether Tesla's autonomous ambitions are outrunning its safety practices — and they land right as the broader robotaxi industry faces fresh Congressional scrutiny. A proposed bill would require federal minimum safety standards for all autonomous vehicles responding to emergencies. More on our transport and mobility coverage.

NASA hired a robotic spacecraft called Link to rescue the Swift space telescope — but the rescue mission itself is now in trouble. Two of Link's three reaction wheels, which control its alignment in orbit, have failed. A third is also showing problems. It's a reminder that space robotics is still genuinely hard, even when the goal is saving another broken spacecraft.

New details have emerged about OpenAI models being used to exploit a zero-day vulnerability in JFrog Artifactory, giving attackers access to Hugging Face's systems. Ten full days passed between the initial exploit and a patch being released — a significant window. It's a sharp reminder that as AI tools become more capable, they're also becoming more useful to attackers. The security angle here deserves more attention than it's been getting.

Following UK staff at the Wikimedia Foundation announcing their intention to unionise last month, US-based employees are now expanding the effort. The nonprofit that runs Wikipedia is becoming an unexpected front in the broader labour movement in tech. It won't reshape the industry on its own, but it signals that even mission-driven organisations aren't immune to the tensions between workers and management that have been building across Silicon Valley.

Between AI workers asking to be regulated and a robot spinning out of control in orbit, today felt like a day where the technology got a little too real for everyone involved — including the people building it.