Richard Gere's Tibet group was a target of the AI spy op Anthropic stopped
Anthropic's biggest espionage takedown just got a famous face, Europe's drone money closed at a $6.4 billion valuation, and the smartest bed you can buy learned some new senses. Three stories, three altitudes.
When Anthropic published its threat-intelligence report earlier this month, the document named its targets. One of them reads like a casting decision — because he is one. In a Wall Street Journal op-ed published today, actor Richard Gere writes that the International Campaign for Tibet, the Washington advocacy group he chairs, was among the organizations surveilled by GTG-14020, the China-based, government-aligned intelligence operation Anthropic says it disrupted. According to the report, the operators used Claude as a stand-in for an analyst team: collecting birth dates, immigration records and social handles on Catholic, Tibetan Buddhist, Falun Gong and Taiwanese Christian communities, mapping religious venues down to floor plans, and filing daily situation reports drawn from WeChat, Xiaohongshu, Douyin and Western platforms. The prompts told the model to adopt "China's standpoint" and to describe Tibet's exile administration as an "illegal separatist administration" — the vocabulary of the United Front Work Department, not a casual user, and one account holder openly described themselves as a Chinese state information-security officer. Anthropic banned the accounts and published indicators of compromise, sharing findings with authorities and industry partners. But targets like Gere learned where they sat in the dataset from a public report, not a phone call — and his op-ed is the first victim going on the record about it.
The attribution here is Anthropic's assessment, not an admission by any state, and there is no evidence yet the surveillance produced offline harm. What is new is the shape of it: a frontier model doing the grunt work of foreign-intelligence collection, and the named targets finding out from a company blog post. Expect that notification question to follow labs for a while.
Europe's defense-AI wave just printed another big number. Tekever, the Portuguese maker of AI-powered surveillance drones, has raised $580 million at a $6.4 billion valuation — the first close of a Series D co-led by UC Investments and Baillie Gifford, per Reuters. The round follows the UK Ministry of Defence's July selection of the company for a surveillance program worth up to £400 million over ten years, and its drones have logged more than 50,000 operational flight hours in Ukraine since 2022. Tekever now sits in the same rearmament current as Helsing and Quantum Systems: AI autonomy is where NATO-adjacent capital is flowing, and a $6.4 billion valuation for a drone maker says investors expect that flow to keep widening.
Eight Sleep's new Pod 6 shrinks the machine and multiplies the senses. The under-bed cooling hub is half the size of its predecessor and adjusts temperature 20% faster, while the mattress cover now packs 18 biometric sensors — up from two — tracking heart rate, breathing and snoring for two sleepers and distinguishing adults from children and pets. Everything feeds Autopilot, the subscription AI that sets your bed's temperature all night without being asked. A new $1,999 Solo option covers half a mattress for single sleepers; full sizes run $2,699 to $2,999, plus $199 to $399 a year for the brains. The AI story is the closed loop — sense, decide, actuate — and it is the clearest sign yet that ambient health computing starts in the bedroom.
What to watch: whether more GTG-14020 targets follow Gere onto the record — and whether Anthropic starts notifying surveilled organizations before publishing their names.
If an American chatbot had been used to profile your organization, would you want a private call or a public report? Tell us in the comments.
