U.S. Treasury Secretary Scott Bessent warned that sanctions are possible after White House officials accused Chinese AI company Moonshot of improperly distilling Anthropic's Fable model to develop its own K3 model. The White House alleged Moonshot obtained Nvidia GB300 servers banned from export to China and conducted large-scale distillation against U.S. AI models. The accusations have intensified debate in Washington about whether the U.S. should restrict Chinese open-weight AI models to preserve technological advantage.
White House official Michael Kratsios accused Chinese AI startup Moonshot of using deceptive methods to extract data from Anthropic's Fable 5 model and use it to train its Kimi K3 system, framing model distillation as technology theft rather than competitive practice. Kratsios alleged Moonshot used a purpose-built platform with multiple access methods to avoid detection and acquired Nvidia GB300 servers in Thailand to train K3, a 2.8 trillion-parameter model released six days after the accusations. The administration's claims could trigger tighter API restrictions, expanded export controls on AI chips, and sanctions on Moonshot, though the allegations lack publicly verifiable evidence and the companies have not commented.
Anthropic is launching a $200 million Economic Futures Research Fund to support external research on how to prepare society for AI's economic impacts, focusing on worker transitions, income support modernization, and ensuring AI gains are broadly shared. The fund will primarily support projects in the $5–30 million range and prioritize five research areas: workplace AI integration, retraining and job placement, AI-driven displacement support, worker stakes in AI growth, and public investment effectiveness. Research findings will be shared publicly at key milestones to help workers, firms, and governments adapt to potential AI-driven economic disruption.
Verona, Wisconsin covered three Flock license plate recognition cameras with trash bags after the city council voted not to renew its contract, because Flock refused to remove them and city officials were uncertain about their legal authority to do so themselves. Flock told the city it was unsure whether the cameras could be remotely disabled, and scheduled maintenance rather than removal work orders. The incident illustrates how cities attempting to end surveillance camera contracts face delays and legal ambiguity when vendors resist cooperating with decommissioning requests.
Anthropic warned that Chinese AI developers are distilling capabilities from American frontier models, but hours later Thinking Machines revealed its new foundation model incorporated Chinese models DeepSeek-V3 and Moonshot AI's Kimi K2.5, exposing the contradiction in U.S. containment efforts. Major AI laboratories worldwide including OpenAI, Anthropic, Google DeepMind, Meta, and Chinese firms all routinely use teacher-student model training and synthetic data generation, meaning the competitive question is now whose models become teachers rather than whether distillation occurs at all. As Chinese frontier models like Qwen and Ernie become embedded in global infrastructure—including Apple's China-specific AI stack using Alibaba and Baidu models—the U.S. faces a policy paradox: tightening controls on its own closed models while confronting openly distributed Chinese alternatives that remain beyond regulatory reach.
Anthropic donated an additional $20 million to Public First Action, a non-partisan organization advocating for AI safeguards and policy, bringing total support to $40 million. The donation supports public education efforts and policy work, with Anthropic citing rapidly advancing AI capabilities—including Claude Mythos Preview discovering thousands of software vulnerabilities—as justification for stricter government oversight, transparency requirements, and safety testing before deployment. The funding aims to build political support for regulations including independent model evaluation, export controls on chips, and government authority to slow or block AI models posing catastrophic risk.
Every AI story that matters,
in your inbox by 8am.
TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the
day in two minutes. Follow companies and topics for alerts, or get the
briefing in Slack. Free, no spam, unsubscribe anytime.
Reading TLDRocket needs no cookies, and the readership counts we rely on come from
our own cookieless analytics. Google Analytics is the exception: it sets cookies and
reports to Google, so it stays switched off until you allow it. You can change your
mind any time from “Cookie settings” in the footer.
Strictly necessary
Session security and form protection (tldrocket-session,
XSRF-TOKEN, 2 hours). The site cannot work without them,
so they need no consent.
Always on
Google Analytics 4 (_ga,
_ga_<id>, up to 2 years). Measures which
stories and sections readers use. Google acts as a third-party processor and may
store the data outside the EU. No advertising, no profiling, no data sold.
