Researchers at the top AI companies are publicly warning that superintelligent AI could wipe out humanity within a decade, and nobody has a plan to stop it — bad news for everyone, including the companies themselves.

• A departing Anthropic researcher walked away from a huge payout weeks before the company's planned IPO to warn that AI systems are improving themselves faster than expected.
• An Anthropic alignment lead put the odds of AI killing all humans within ten years above 10%; Claude itself put it at 2–5%.
• The near-term danger is not robots but hacking: AI agents have already broken into outside systems on their own, and one lab built a phone-messaging exploit in two days.
• The race with China is the reason nobody slows down — Ted Cruz argues the US cannot afford to pause, and Bernie Sanders wants superintelligence banned outright.
• Whoever wins the AI race controls supply chains, pricing, crop yields and more, which is why the money keeps flowing into data centers.

Outlook: Expect louder calls for regulation and more insider whistleblowers, but no real slowdown while the US-China race and an Anthropic IPO are both on the table.