AI News for May 15, 2026 — 7min.ai

[NEWS] Cerebras pops 68% on debut, hits $100B market cap: Cerebras Systems opened at $350 a share Thursday, nearly double its $185 IPO price, then settled at $311.07 for a 68% day-one gain.
[NEWS] OpenAI preparing legal action against Apple over Siri partnership: OpenAI has hired outside counsel to weigh a breach-of-contract notice or lawsuit against Apple over the ChatGPT-in-Siri deal struck at WWDC 2024.
[NEWS] Musk v. Altman jury deliberates after chaotic closing arguments: Nine California jurors began deliberating Thursday after closing arguments in Musk v.
[TOOLS] OpenAI brings Codex to the ChatGPT mobile app: OpenAI launched a Codex tab inside the ChatGPT app for iOS and Android, letting users start, monitor, and approve coding tasks running on their desktop from a phone.
[RESEARCH] Ontario audit: 20 AI medical scribes hallucinate and miss key details: Ontario's Auditor General tested 20 AI medical scribe vendors pre-approved by the provincial government and found every single one had accuracy problems.
[RESEARCH] Anthropic Mythos cracks a 5-year macOS defense in five days: A California security team published details on how Claude Mythos Preview helped them build the first public macOS kernel memory corruption exploit on M5 silicon, bypassing a five-year Apple security effort in five days.
[NEWS] China blocks 10 cleared firms from buying Nvidia H200 chips: The US cleared roughly 10 Chinese companies, including Alibaba, Tencent, ByteDance, and JD.com, to buy Nvidia H200 chips at up to 75,000 units each, with Lenovo and Foxconn approved as distribution partners.
[NEWS] Microsoft cancels most Claude Code licenses, pushes Copilot CLI: Microsoft is pulling most internal Claude Code licenses from its Experiences + Devices org (Windows, Microsoft 365, Outlook, Teams, Surface) by June 30, the last day of its fiscal year.
[RESEARCH] Microsoft's MDASH agents find 16 Windows vulnerabilities, 4 critical: Microsoft revealed MDASH (Multi-Model Agentic Scanning Harness), a system orchestrating more than 100 specialized AI agents across frontier and distilled models to hunt software vulnerabilities.
[RESEARCH] Claude Mythos clears every UK AI safety cyberattack simulation: The UK AI Security Institute revised its forecast for AI cyber capability doubling time from 8 months to 4.7 months, and says Claude Mythos Preview and GPT-5.5 have already blown past even that curve.

About 7min.ai · Subscribe to newsletter