AI News for October 31, 2026 — 7min.ai

[NEWS] Official Claude Code extension backdoored on Cursor's marketplace for nearly 12 hours: A malicious version (2.1.313) of the official Claude Code extension sat on Cursor's marketplace for nearly 12 hours, reaching at most 5,000 users.
[NEWS] Intel plunges 28.6%, its worst day since 1974, as AI memory crunch hits PCs: Intel fell 28.6% on Friday, its worst drop since 1974, after guiding Q4 revenue to $15-16B against expectations near $17B.
[NEWS] AI selloff knocks nearly $1T off the Nasdaq in a week: The Nasdaq lost nearly $1T in a week and about $2T in a month as investors questioned whether AI spending will pay off.
[NEWS] OpenAI raises less than half of $30B target as AI selloff bites: OpenAI cut its fundraising target by $10B and has raised less than half of the $30B it sought in its latest round.
[RESEARCH] Open-weight GLM 5.3 builds Firefox exploits for about $1,000 with no human input: Epoch AI showed that Z.ai's open-weight GLM 5.3 can autonomously attack Firefox in the "patch gap," when a fix is visible in code but hasn't reached users.
[NEWS] Anthropic may get a 'junk' credit rating after its IPO, Blackstone tells investors: Blackstone told investors it expects Anthropic to land a double-B rating after its IPO, the top tier of speculative "junk" grade.
[NEWS] OpenAI, Google, Microsoft and Anthropic back shared robots.txt opt-out for AI training: OpenAI, Google, Microsoft and Anthropic will honor a shared "policy token" in robots.txt that lets publishers set AI training preferences once, instead of company by company.
[RESEARCH] Google's Suncatcher satellite runs Gemma 4 in orbit and sends back a haiku: Google's Project Suncatcher satellite booted its TPUs and returned its first AI response from space, a haiku from an on-board Gemma 4 model.
[RESEARCH] A Claude agent chose a gene therapy and operated on a mouse brain: An agent running Claude Sonnet 5 picked a gene therapy from 19 options, chose the brain region and coordinates, then ran the injection through a motorized rig on an epilepsy mouse model.
[RESEARCH] Anthropic has Claude flag misuse in its own conversations: Anthropic detailed "self-assessment," in which Claude reviews a chat after each reply and gives a safety verdict covering harm, type and urgency.

About 7min.ai · Subscribe to newsletter