AI INTELLIGENCE / DAILY SIGNAL
0011011011
2026-07-11 / AI NEWS

AI News Intelligence Digest — July 11, 2026

Apple sues OpenAI for trade secret theft — blockbuster reversal of their 2024 partnership. Apple filed in Northern California federal court alleging OpenAI's hardware chief Tang Tan directed Apple employees to bring "act...

BRIEFING STRUCTURE24 SIGNALS

Inside this digest

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

Top Stories

  • Apple sues OpenAI for trade secret theft — blockbuster reversal of their 2024 partnership. Apple filed in Northern California federal court alleging OpenAI's hardware chief Tang Tan directed Apple employees to bring "actual parts" to interviews and coached departing staff to evade security. Apple claims 400+ ex-employees now at OpenAI, and the scheme operated "at every level." The trigger: OpenAI's $6.4B acquisition of Jony Ive's IO Products put them on a collision course with Apple's hardware business. OpenAI denied the allegations. This is the biggest inter-company AI lawsuit since Google v. Uber in 2017. (CNBC — "Apple sues OpenAI alleging trade secret theft"; Bloomberg; TechCrunch)

  • SK Hynix lands on Nasdaq at $1.27T — largest foreign IPO in history. The South Korean memory giant raised $26.5B selling 177.9M ADRs at $149, then closed day one up 13% at $168.01. The offering was 7x oversubscribed. SK Hynix controls 60% of the global HBM market — the memory chips NVIDIA can't get enough of. This is a direct play on the AI infrastructure bottleneck: NVIDIA's Jensen Huang says HBM shortages will persist "for quite a few years." SKHY ticker, with leveraged ETFs SKUU/SKDD launching July 13. (Bloomberg; Money Morning; CNBC)

  • Fable 5 billing cliff hits tomorrow — $10/$50 per Mtok kicks in July 13. Developing story, tracked since July 4: Anthropic's extended inclusion window expires July 12 at 11:59 PM PT. After that, every Fable 5 call goes to metered usage credits at 2x Opus 4.8 pricing. Batch API halves it to $5/$25, cache hits drop to $1/Mtok. The question now: will Anthropic bring subscription inclusion back "when capacity allows," or is this the permanent pricing? (Forbes — "Claude Fable 5 Extends By Five More Days"; Digital Applied)

  • Fed taps Marc Andreessen to co-lead AI jobs task force. Fed Chair Kevin Warsh named the a16z co-founder alongside Stanford's Charles I. Jones and Microsoft's Asha Sharma to his Productivity and Jobs panel — one of five external task forces conducting a sweeping monetary policy review. Recommendations due end of 2026. The optics: the VC who's bet billions on AI now advising the Fed on whether AI will eliminate jobs. (Washington Post; Axios; CNBC)

  • Japan commits $6.7B to sovereign AI — SoftBank, Sony, Honda, NEC consortium. Noetra, the newly formed joint venture, will build physical AI foundation models trained on Japanese industrial data — not chatbots. Honda contributing 2.3M hours of driving data, Sony opening CMOS sensor datasets, NEC sharing biometric corpora. First model this fiscal year. This is Japan saying it won't depend on Silicon Valley for the AI that runs its factories and robots. (Asia Times; Japan Times; Finimize)

Model Updates

  • GPT-5.6 settling in post-launch. Luna ($1/$6) emerging as the developer default for routine work, Sol Ultra for complex agentic tasks. Sol on Cerebras hitting 750 tok/s. Atlas browser retiring August 9 — features folding into ChatGPT desktop app. (Android Authority; The Register)
  • Gemini 3.5 Pro reportedly targeting July 17 GA. Full architectural rebuild with 2M-token context window and Deep Think reasoning. Six weeks late from the original June target. Google scrapped the old base rather than iterating. Unconfirmed until Google publishes model card. (BigGo Finance; Geeky Gadgets)
  • Grok 4.5 hallucination problem persists. Artificial Analysis confirms 54% hallucination rate (up from 25%) even as accuracy improved. MoE architecture learns confidence including about wrong answers. At $2/$6 per Mtok it's cheap, but the accuracy tradeoff limits high-stakes use. EU launch still pending AI Act compliance. (TechTimes; Artificial Analysis)
  • Muse Spark 1.1 in public preview. Meta's first paid API model at $1.25/$4.25 per Mtok. 1M token context with active context management. Strong at tool use and coding. (Fortune; TechCrunch)

Open Source

  • GLM-5.2 remains the open-weight coding king — 62.1% SWE-bench Pro, MIT license, 27x daily token growth since Vercel launch. Still beating GPT-5.5's 58.6% on that benchmark. (Thunder Compute; Taskade)
  • Kimi K2.7 Code shows strong agentic coding — +21.8% over K2.6 on Kimi Code Bench v2. The Chinese open-weight wave continues. (LLM Stats)

Industry Moves

  • Apple v. OpenAI could reshape AI talent movement if Apple wins injunctions against former employees. Watch for OpenAI's response filing.
  • SK Hynix IPO validates AI infrastructure as its own investment thesis — 72% operating margins on HBM chips.
  • SpaceXAI leasing Colossus compute to Anthropic ($1.25B/mo) and Google ($920M/mo) while training Grok on the same cluster. Structural tension.
  • Together AI raised $800M Series C led by Aramco Ventures. AI inference infrastructure remains a capital magnet.
  • HHS deploying ChatGPT for state Medicare/Medicaid audits across all 50 states — first major federal operational deployment.
  • Three humanoid robot companies heading to public markets simultaneously: Agility ($2.5B SPAC), Unitree (Shanghai STAR Market), Tesla ramping Optimus to 1,000/week by September.

Safety & Policy

  • Illinois SB 315 signed July 6 — nation's first frontier-model-specific safety law. Requires annual independent third-party audits. Both OpenAI and Anthropic supported it. Takes effect January 1, 2027. CA+NY+IL together cover ~40% of US AI market.
  • UN Geneva AI Governance Dialogue (July 6-7): Yoshua Bengio told governments "science currently cannot guarantee AI will not cause catastrophic harm." 99% of deepfakes are sexual, 96% target women.
  • China's Anthropomorphic AI rules take effect July 15 — covering AI that simulates a person for ongoing emotional interaction.
  • FTC proposed policy statement (July 1) targeting AI companies that manipulate outputs in ways violating consumer objectivity expectations.

Worth Watching

  • "AI News in a Minute | Thursday, July 9, 2026 Episode 1" — Sharp breakdown of Grok 4.5's pricing disruption and the Cursor-free-tier angle.
  • "Exciting AI Updates Weekly - July 10, 2026" by Lev Selector — comprehensive weekly roundup covering the GPT-5.6 launch aftermath and market reactions.
  • "MUST-SEE Which AI Companies Are Winning | AI Daily News July 07, 2026" — good market analysis of the competitive landscape heading into the busiest model-launch week in AI history.