AI INTELLIGENCE / DAILY SIGNAL
0011011011
2026-05-18 / AI NEWS

AI News Intelligence Digest — May 18, 2026

Musk Loses OpenAI Trial — Jury Takes Less Than Two Hours. The nine-person advisory jury unanimously rejected all of Elon Musk's claims against Sam Altman and OpenAI today, deliberating for under two hours. The decisive b...

BRIEFING STRUCTURE22 SIGNALS

Inside this digest

  • Top Stories
  • Model Updates
  • Open Source
  • Industry Moves
  • Safety & Policy
  • Worth Watching
← Back to archive

Top Stories

  • Musk Loses OpenAI Trial — Jury Takes Less Than Two Hours. The nine-person advisory jury unanimously rejected all of Elon Musk's claims against Sam Altman and OpenAI today, deliberating for under two hours. The decisive blow: statute of limitations — the jury found Musk knew about OpenAI's for-profit shift as early as 2021 but didn't file until 2024. Judge Gonzalez Rogers agreed, dismissing breach of charitable trust and unjust enrichment claims as untimely. Musk's $134B remedies demand, leadership removal, and for-profit unwinding — all dead. Musk called the judge a "terrible activist" on X. His attorney Marc Toberoff confirmed they will appeal. OpenAI attorney William Savitt: "a hypocritical attempt to sabotage a competitor." The practical impact: OpenAI's IPO path is now clear of its biggest legal overhang. Resolution — tracked since May 15. Appeal likely but the core threat is neutralized. (CNBC, CNN, TechCrunch, Newsweek)

  • Google I/O Tomorrow: The AI Industry's Biggest Stage This Week. Keynote opens May 19, 10am PT at Shoreline Amphitheatre. The entire two-hour opener is AI. Expected: new Gemini model (leaked ARC-AGI-2: 84.6% vs Opus 4.6's 68.8%), Gemini Spark autonomous agent (24/7, creates reusable skills, may make purchases without asking), Android XR glasses preview, Android 17 unifying Android/Chrome OS/XR, and Googlebooks — AI-native laptops from Acer, Asus, Dell, HP, Lenovo shipping this fall. No Pixel hardware. This is Google's chance to prove it hasn't fallen behind Claude Mythos and GPT-5.5. Developing — tracked since May 15. Tomorrow is confirmation day. (Android Authority, TechTimes, Gizmodo)

  • OpenAI and Dell Partner to Bring Codex On-Premises. Announced today at Dell Technologies World: Codex integrates with Dell AI Data Platform and Dell AI Factory for hybrid/on-premises enterprise deployment. 4 million developers now use Codex weekly. Codex agents expanding beyond coding — preparing reports, routing feedback, qualifying leads, coordinating cross-system workflows. This is significant: enterprise AI moving from cloud-only to on-prem, meeting data sovereignty and security requirements where they live. Google Gemini and Palantir Foundry also coming to Dell infrastructure. (OpenAI, SiliconANGLE, Dell)

  • Meta's 8,000 Layoffs: T-Minus 2 Days. Cuts begin May 20 — company-wide and structural, not performance-based. 8,000 employees plus 6,000 canceled reqs (14,000 effective headcount reduction). Teams reorganized into AI-focused "pods" under Alexandr Wang's Superintelligence Labs. $115-135B AI capex this year. More cuts planned for H2. Total since 2022: ~25,000 — despite $201B revenue (up 22% YoY) and $43.6B free cash flow. Developing — tracked since May 14. Execution begins Tuesday. (TheNextWeb, CNN, Quartz)

  • Cerebras Week 1 Ends Ugly: CBRS at $262 After Hours. Closed at $279.72 (-3.95%), then fell further to $262.50 after hours (-6.16%). Now down 32% from the $386 intraday peak on IPO day. The $275 support level we flagged? Broken intra-day. Still 42% above $185 IPO price, but the post-IPO honeymoon is definitively over. Classic volatility pattern continues. Developing — tracked since May 14. (Yahoo Finance, Investing.com)

Model Updates

  • Claude for Small Business launched — integrates with QuickBooks, PayPal, HubSpot, Canva, Docusign, Google Workspace, Microsoft 365. Ready-to-run workflows for payroll, invoicing, sales, marketing. (Anthropic)
  • Claude Opus 4.7 Fast Mode now in research preview — faster output token generation on the developer platform. (Anthropic)
  • Meta Muse Spark (née Avocado) — rebranded and released from Meta Superintelligence Labs. First model in the new Muse series. Closed-source, a break from the Llama open-source tradition. Internal benchmarks placed it between Gemini 2.5 and 3.0. (CNBC, TechBuzz)
  • Zhipu AI GLM-4.7 — trained entirely on Huawei Ascend silicon (no NVIDIA). Claims 1.2% hallucination rate (lowest reported by any frontier lab). $0.11/M input tokens vs Claude Opus at $15. (MLQ)

Open Source

  • NVIDIA Cosmos physical AI models — Cosmos Reason 2 (robot reasoning VLM), Transfer 2.5 and Predict 2.5 (synthetic video generation). Open-source, 2M+ downloads. 10T language tokens, 500K robotics trajectories, 100TB vehicle sensor data contributed. (NVIDIA)
  • NVIDIA Ising — open-source AI models for quantum computing error correction. 2.5x faster, 3x more accurate than traditional approaches. (NVIDIA)
  • Mistral 3 continues gaining momentum — GitHub forks doubled, merged PRs tripled in three months. Open-source quality converging with proprietary. (Various)

Industry Moves

  • Anthropic approaching $1T valuation — in talks for $50B round at ~$900B. Revenue exploded to $30B+ annualized. Goldman Sachs, Blackstone, Hellman & Friedman in $1.5B side venture. Claude Code now authors 4% of public GitHub commits. (TechCrunch, TechFundingNews)
  • Dell Technologies World — Dell Deskside Agentic AI (secure local sandbox for AI agents on workstations), Dell PowerRack (rack-scale AI, delivery to production in 6.5 hours). (IT Pro, SiliconANGLE)
  • Shield AI — $1.5B Series G at $12.7B valuation (up 140% in one year). Defense AI. (Crunchbase)
  • Rhoda AI — $450M Series A for robotic intelligence platform FutureVision. (Crunchbase)

Safety & Policy

  • EU AI Act Omnibus agreed May 7 — simplifies compliance, extends high-risk AI deadlines to Dec 2027 (Annex III) and Aug 2028 (Annex I). SME exemptions expanded to companies up to 500 employees. New prohibition on "nudifier" AI apps effective Dec 2, 2026. Formal adoption expected before Aug 2, 2026. (EU Council, White & Case)
  • US regulatory patchwork intensifies — 1,561 AI bills introduced across 45 states as of March, surpassing all of 2024. White House wants federal preemption but Congress stripped a proposed 10-year freeze from the "One Big Beautiful Bill Act" by 99-1 vote. No unified federal framework in sight. (Fortune, Ropes & Gray)
  • Frontier labs opening to regulators — Microsoft and xAI agreed to provide early model access to regulators, a meaningful shift toward voluntary transparency. (MarketingProfs)

Worth Watching