Anthropic is signing users out, and shrinking Claude Code's limits
Anthropic began contacting Claude users on Sunday whose session cookies were stolen by infostealer malware, signing them out and stripping saved payment methods, and separately confirmed that Claude Code's weekly limits will be 17 per cent lower from 14 September than they are today.
Anthropic is signing out Claude users whose sessions malware stole Anthropic began contacting affected users on Sunday after infostealer malware on their own machines copied active Claude session cookies, letting attackers replay a logged-in session and burn through paid usage without ever needing a password or clearing two-factor authentication. It has signed the stolen sessions out, removed saved payment methods and is refunding charges it identifies as unauthorised; the families named are Vidar, LummaC2, StealC, RedLine and Acreed on Windows, plus Atomic Stealer on a small number of Macs. Source
A lawsuit says xAI trained Grok on child sexual abuse material A class action in the Northern District of California, which an anonymous survivor joined last week as a lead plaintiff, alleges that images of her childhood abuse were in the data used to train xAI’s image and video generation. It also alleges that xAI treats Grok’s own output as training-eligible, so abuse material the model produced was fed back into it, and it asks the court to order that material destroyed. xAI has not publicly responded to the training-data allegation. Source
Claude Code’s weekly limits drop 17 per cent on 14 September Anthropic said at the weekend that standard weekly limits on Pro, Max, Team and seat-based Enterprise plans rise permanently by 25 per cent from 14 September, but a temporary 50 per cent boost ends on 13 September, so against what users have right now it is a cut. The company deleted its original post after developers pointed that out and reposted one saying: “Compared to today, this works out to a 17% reduction in weekly limits on Claude Code.” Source
Chatbots handled state propaganda better than AI search summaries did NPR and NewsGuard put 30 questions built from false narratives pushed by Russia, China and Iran between December 2025 and July 2026 to the major chatbots, including ChatGPT and Gemini, and published the results on Sunday: the chatbots mostly pushed back. The AI summaries sitting above the results on the big search engines did not do as well. Source
John Ternus takes over as Apple’s chief executive today Tim Cook becomes executive chairman and Ternus, until now Apple’s head of hardware engineering, becomes chief executive from today, under a succession the company announced in April. Bloomberg’s Sunday piece on the handover puts AI at the top of his in-tray. Source
Claude Sonnet 5’s price did not go up today after all The $2 and $10 per million input and output tokens Claude Sonnet 5 launched with in June were sold as introductory pricing through 31 August, with a rise to $3 and $15 due today. Anthropic made the introductory rate permanent on 10 August, and its pricing page now says plainly that the increase will not occur — worth knowing if a roundup told you otherwise this morning. Source