Apple CEO John Ternus calls memory chip shortage a crisis

Meta recently revised its commercial terms, removing the word "train" after criticism that the language would let the company scrape publisher sites for AI training. The change, effective Oct. 30, still allows content collection for advertising products, leaving publishing groups uneasy. Meanwhile, Replit announced that OpenAI’s models are now the default on its platform, citing cost and performance advantages, though users can still select Anthropic models. Anthropic released an internal report showing its junior technical staff share dropped from 21% to 14% and new junior hires fell from 27% to 9%, while 27% of employees said Claude boosted productivity.

Google co‑founder Sergey Brin hinted at "recursive self‑improvement" for Gemini 4 Argon, while a DeepMind safety report downplayed rapid AI‑driven research acceleration. Bill Gates is reportedly expanding his AI‑safety outreach, planning direct talks with CEOs. These developments illustrate the tug‑of‑war between optimism in AI hardware demand and caution over market bubbles and workforce impacts.