AI News for May 9, 2026 — 7min.ai

[NEWS] Anthropic signs $1.8B Akamai cloud deal as 80x growth strains compute: Anthropic inked a seven-year, $1.8 billion compute deal with Akamai, adding to its earlier SpaceX Colossus and $200B Google Cloud commitments.
[NEWS] DeepSeek nears $50B valuation in $7.35B raise, plans V4.1 for June: DeepSeek is closing its first external funding round at up to 50 billion yuan ($7.35B), the largest ever for a Chinese AI company, lifting its valuation to roughly $50 billion.
[NEWS] Intel jumps 19% on preliminary Apple chip deal, stock up 240% YTD: Intel and Apple reached a preliminary agreement for Intel to manufacture some chips for Apple devices, sending Intel shares up as much as 19% Friday.
[NEWS] Federal court rules DOGE's 'is this DEI?' ChatGPT prompt unconstitutional: US District Judge Colleen McMahon ruled DOGE's cancellation of more than $100 million in National Endowment for the Humanities grants unconstitutional, citing how a DOGE staffer fed each grant description to ChatGPT with the prompt: "Does the following relate at all to DEI?
[NEWS] ByteDance lifts 2026 AI capex 25% to $30B as memory costs climb: ByteDance is raising 2026 capital expenditure on AI infrastructure to more than 200 billion yuan ($30B), up at least 25% from a preliminary plan of 160 billion yuan discussed late last year.
[RESEARCH] Anthropic's 'Teaching Claude Why' drives blackmail rate to zero post-training: Anthropic published research detailing how it eliminated agentic-misalignment behaviors after Claude 4, where models had blackmailed engineers up to 96% of the time in test scenarios.
[TOOLS] OpenAI launches GPT-5.5-Cyber tier with relaxed filters for vetted defenders: OpenAI released GPT-5.5-Cyber through its Trusted Access for Cyber program, offering authorized critical-infrastructure defenders a model variant with the fewest content restrictions for penetration testing and malware analysis.
[NEWS] RedAccess: 5,000 vibe-coded apps leak corporate data on the open web: Israeli security firm RedAccess found 380,000 publicly accessible apps and databases built with Lovable, Base44, Replit, and Netlify, with about 5,000 (1.3%) exposing sensitive corporate data.
[NEWS] Google's Isomorphic Labs to raise over $2B for AI drug discovery: Isomorphic Labs, the AI drug-discovery company spun out of Google DeepMind, is in advanced talks to raise more than $2 billion.
[NEWS] NERC issues rare Level 3 alert as AI data center power swings threaten grid: The North American Electric Reliability Corporation issued a Level 3 "essential action" alert, its highest tier, warning that grid operators lack adequate processes to handle the rapid power swings from data centers running AI workloads.

About 7min.ai · Subscribe to newsletter