WIRED Roundup: AI Psychosis, Missing FTC Files, and Google Bedbugs (www.wired.com)

WIRED’s Uncanny Valley roundup covers several AI-related shifts: retailers are already preparing for a holiday surge in chatbot-driven shopping as AI search/chat interfaces replace traditional web search, prompting a move from SEO to “generative engine optimization” (GEO). OpenAI’s Walmart partnership to enable in-chat purchases underscores that change; one GEO consultant says overlap between top Google links and sources cited by chatbots has dropped from roughly 70% to under 20%, meaning companies must restructure content (clear use-cases, bulleted facts, richer product metadata) to surface in LLM answers. Separately, multiple people have filed complaints with the FTC claiming ChatGPT contributed to so‑called “AI psychosis,” raising fresh questions about mental‑health harms, explainability, and where liability and consumer protection should sit. Those user harms sit alongside growing regulatory uncertainty: the FTC has removed several blog posts published during Lina Khan’s tenure — including discussions of open‑weight models (publicly released, inspectable/modifiable models) and consumer AI risks — and previously took down hundreds of AI-related posts, creating opacity about agency guidance. For the AI/ML community this matters technically and operationally: model transparency (open‑weight vs. closed models), provenance and citation of training sources, safety mitigations for psychological harms, and evolving compliance expectations will all influence model design, deployment practices, dataset curation, and how companies surface information to LLMs.