World Liberty Financial is backing WorldClaw, an AI platform that offers AI models from Chinese companies flagged by the U.S. for national security and intellectual property concerns.
The collaboration involves Chinese firms restricted by the U.S. government.
This creates a backlash over whether a U.S.-linked crypto venture is effectively routing around government restrictions to profit from the underlying AI tools.
A hidden Airtag planted in a rare book from a bulk order was tracked to an Amazon AI training facility in Las Vegas where books were torn from their spines and scanned. The Airtag led to the VGT3 warehouse associated with a team that processes and destroys books, according to the report. The finding strengthens evidence that Amazon is involved in destructive book scanning for AI training, amid backlash.
Anthropic CEO Dario Amodei argued that open-weights only shifts AI power to whoever controls the most compute and chips, not to truly independent developers. He cited California’s SB 53 definition of a frontier model as one trained using more than 10^26 FLOPs and requiring extra rules for developers earning over $500 million per year. He said regulation should be tiered (frontier vs off-frontier) and linked to safety testing, while still leaving room for smaller labs rather than enabling incumbents or banning open models wholesale.
Simon Willison’s Weblog·2 months ago·
24
● 7 sources
404 Media tracked a shipment of rare books from a dealer to an Amazon AI training facility after suspected AI-scan bulk orders. The bookseller received an order of around 1,000 books in July. The reporting links the shipment to a Las Vegas Amazon site (VGT3) where workers said books are destructively scanned for AI training, strengthening evidence of such practices.
An expert witness hired by 3M used ChatGPT to draft parts of his expert report in a Houston explosion lawsuit tied to the 2020 Watson Grinding disaster. The record shows 3M paid about $90,000 for the analysis at a rate of $475 per hour, and the prompts sought to “show how 3M is 0% at fault.” The case exposes that ChatGPT prompts and conversations can be discoverable in litigation, and this changed the trial focus as attorneys used the AI use to challenge the report’s credibility.
A 404 Media investigation tracked a rare-book shipment with a device to identify an AI buyer and found it ending at an Amazon facility that scans and destroys books for AI training data. The report cites Amazon’s VGT3 symbol for the Las Vegas facility involved. As a result, the story says Amazon’s previously unreported scale of buying books for scanning and destruction is now documented.
Semafor editor-in-chief Ben Smith discussed the ethics and business model behind Semafor’s journalism as audiences shift away from legacy media. Semafor launched in 2022 with a goal of reaching 200 million college-educated readers. The company describes itself as an anti-scale, no-website-paywall outlet that earns about half its revenue through convening (events) and frames its approach to AI and editorial independence around those ethical tradeoffs.
The FCC added “advanced robotic devices” to its national-security ban list, blocking new foreign-made humanoids, quadrupeds, robot vacuums, and lawn mowers from entering the U.S. if they are assembled in the U.S. with at least 65% of components by value produced domestically. The domestic-parts requirement will increase to 75% in 2029, which is stranding robotics startups that relied on China’s electronics supply chain for prototyping and component sourcing. Startups say they must restructure manufacturing plans and supply sources, while some research and software teams may shift purchasing toward U.S.-made robots instead.
Anthropic CEO Dario Amodei denied claims that he wants Anthropic to become the only company left standing in AI. His two-part reply drew 10 million views within a day. The dispute pivoted to his argument that regulation should create accountability based on trust and transparency rather than leaving power to a few giant firms.
Anthropic clarified that Claude-generated text will include invisible watermarking to meet Europe’s AI transparency rules. It said the watermarking system will use a SynthID-Text approach from Google DeepMind. As a result, Claude will add machine-detectable marks to text and also add C2PA support for images it processes.
OpenAI started emailing users in the EEA and Switzerland that it will add ads to the Free and Go versions of ChatGPT later this month. The Go plan will cost €7.99 per month while still including ads. Ads are still non-personalised at launch, with personalisation requiring an opt-in, and ad data deletion is available via Settings > Ad controls.
OpenAI funded 14 independent projects exploring new AI policy ideas to expand economic opportunity and strengthen societal resilience. The funding covers 14 projects. This changes the policy debate by adding new, externally developed research contributions to guide work on AI governance and resilience.
Dario Amodei said AI is facing a crisis of trust in which people suspect companies and governments will “screw them over,” even as he pushed back on claims that he drives public doom. He cited his earlier 50% white-collar job wipeout prediction as context while saying he now describes AI as a multiplier of output rather than a destroyer of jobs. As a result, Anthropic is expected to focus more on delivering concrete results in biology and medicine while avoiding empty promises and emphasizing honest risk discussion.
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.
