Issue #045 — 2026-10-06

The Pentagon says it has stopped using Anthropic's AI tools, a defense official told the BBC. People familiar say Claude was still used for intelligence work and in military operations against Iran as recently as last week. Defense Secretary Pete Hegseth labeled Anthropic a supply chain risk in February and set a late August cutoff.

McDonald's was sued in federal court in Chicago in a proposed nationwide class action that says its AI-powered pricing system lets it fix menu prices across franchises, Reuters reports. The suit says the "coercive price-fixing agreement" has run since 2019 and could cover millions of customers. McDonald's says AI does not set the price of a Big Mac.

A federal judge in Mississippi, Henry Wingate, issued a restraining order that cited parties and sworn statements that did not exist, and a law clerk's AI tool produced the errors, per Virginia Lawyers Weekly. He pulled the order and replaced it. A 5th Circuit panel has since heard arguments on whether to remove him from the case, and Senator Chuck Grassley asked for an explanation.

Software engineer Chris Dzoba, who is not a mathematician, had Anthropic's Claude Fable and OpenAI's GPT-6 Astra work together on a message board and found a 17-set symmetrical Venn diagram in two days, per New Scientist. The last record was 13 sets, found in 2014. A 19-set diagram followed that Sunday.

Stanford researchers ran a virtual biotech staffed by 37,000 AI agents, which combed past drug trials and found that drugs aimed at genes active in only a few cell types were more likely to succeed and had fewer side effects, per Earth.com. The software was built in James Zou's lab. Drugmakers could use the clue to pick targets before a trial starts.