The Guardian

Scammers Hijack ChatGPT Sources to Promote Fake Shops

UK consumer-protection groups and fraud detection specialists have identified a new scam where criminals manipulate AI-driven search and recommendation sources to surface fake online stores inside LLM-based shopping answers. The Guardian reports fraud-detection service Ask Silver found forged sites imitating retailers such as Russell & Bromley and Dunelm appearing as sources in ChatGPT search results. These shops offer large discounts, take prepayments and then fail to deliver while harvesting payment and personal data. The phenomenon is described as "AI Poisoning" and experts warn it will spread to other markets. OpenAI has reportedly removed the identified malicious websites from ChatGPT’s search index, but publishers and retailers are urged to prepare for further agentic-commerce abuse.

UK Foreign Secretary Warns of 'AI Hiroshima' Risk

British Foreign Secretary Yvette Cooper warned that the rapid, unpredictable development of artificial intelligence poses an acute danger to humanity and called for urgent international rules to limit AI's destructive potential. In a guest essay for Chatham House, reported by dpa and cited in UK media, Cooper compared the need for AI regulation to the post‑Hiroshima arms-control moment and said powerful states such as the US and China bear special responsibility. She predicted AI will become a dominant foreign‑policy issue within two years and noted non‑state actors and malign governments are already using advanced AI tools for hybrid threats, terrorism, crime and information manipulation.

States Target 'Surveillance Pricing' for Subscriptions

U.S. lawmakers and regulators are moving to curb so‑called “surveillance pricing,” the practice of using personal data to set individualized subscription prices. A June 11 class action accuses The Washington Post of collecting reader data to offer renewal prices between $60 and $170 without disclosure. New York’s legislature passed the One Fair Price Act on June 4 — which, if signed by Governor Kathy Hochul, would broadly prohibit using personally identifiable data (first- or third-party) to set prices and require disclosures for frequent dynamic pricing — joining Maryland and Connecticut, which passed similar laws earlier in the year. The FTC has been targeting surveillance/algorithmic pricing since 2024. Industry lawyers, publishers and trade groups disagree on the impact and scope; the bill allows uniform bona fide discounts and category-based promotions but threatens civil penalties for violations.