California Rejects DC, Adopts Claude.

Mazen Al Sabbagh

— 5 min read
Share

TL;DR

  • California deployment: A precedent-setting agreement integrates Anthropic's Claude across California state agencies, with substantial economic terms, contrasting federal reticence.
  • Seoul's national strategy: South Korea commits over $576 billion to a comprehensive AI and semiconductor initiative, leveraging Samsung and SK Hynix to exceed $1 trillion with data center investments.
  • Anthropic's bifurcated access: Claude Mythos 5 is cleared for critical infrastructure defense, while the public-facing Claude Fable 5 remains federally suspended on day 17.
  • GPT-5.6's restricted launch: OpenAI introduced its new GPT-5.6 tiers to approximately 20 government-vetted partners, reflecting a controlled rollout.
  • Alphabet's Dow entry: Shares rose 4% as Alphabet joined the Dow Jones Industrial Average, pushing the index above 52,000 for the first time.

Lead Story: California Rejects DC, Adopts Claude.

Governor Newsom's administration finalized a strategic partnership with Anthropic this Monday, enabling statewide deployment of Claude at a 50% reduced rate. This agreement, extended to all state agencies and optional for local entities, includes complimentary workforce training and direct engineering support, marking Claude's integration as the inaugural AI tool via the Department of Technology's SITeS shared-services portal. Initial deployments are targeting critical public services, with the DMV utilizing Claude for wait time reduction and the Department of Health Care Services embedding it within internal workflows.

"AI should not replace the human work of government; it should help our workers move faster, solve problems more effectively, and deliver better results for Californians," Newsom stated, framing the initiative as an augmentation of public sector efficiency.

The strategic implications are clear. Washington has designated Anthropic a supply-chain risk and maintained its federal litigation, while Claude Fable 5 has been under federal suspension for over two weeks. Yet, California, the nation’s largest state economy, has designated Claude its default AI assistant. When queried about the federal designation, California's CIO noted to TechCrunch that it "just didn't come up" during negotiations. CBS Sacramento observed this agreement extends Newsom's prior executive order to independently accelerate state AI adoption.

In Other News

Seoul's trillion-dollar AI pivot. South Korea's President Lee Jae Myung, alongside the leaders of Samsung and SK Hynix, unveiled a strategic investment exceeding $576 billion into AI and semiconductor infrastructure. This initiative, centered on a "triple axis" of advanced chips, physical AI, and data centers, commits 800 trillion won ($518 billion) from Samsung and SK Hynix to construct four new fab complexes and allocate $52.5 billion for a specialized chip-packaging cluster. A parallel 550-trillion-won ($356 billion) data-center program, supported by SK Group, GS Group, and Naver, elevates the total national commitment to over $1 trillion. CNN Japan Times Al Jazeera

Mythos 5 cleared for defense; Fable 5 remains sidelined. Commerce Secretary Howard Lutnick's June 26 directive granted Anthropic permission to redeploy Claude Mythos 5—its most robust cybersecurity model—to approximately 100 US organizations engaged in critical infrastructure defense, waiving license fees for defensive applications. CNN Meanwhile, the general-purpose Claude Fable 5 remains suspended globally under the June 12 directive, with penalties still enforced after more than two weeks offline. TechTimes Subscribers and API clients continue to operate on fallback models Opus 4.8, Sonnet 4.6, and Haiku 4.5.

OpenAI's GPT-5.6 debuts with limited access. OpenAI introduced its new three-tiered GPT-5.6 family—Sol (flagship), Terra, and Luna—this Friday, with initial access restricted to roughly 20 government-vetted partners following a federal safety review. Axios Sol, operating in its subagent-driven Ultra mode, achieved a 91.9% score on Terminal-Bench 2.1. VentureBeat OpenAI, in its own announcement, expressed reservations about the process: "We don't believe this kind of government access process should become the long-term default." A broader rollout is anticipated within weeks.

Alphabet enters the Dow; index surpasses 52,000. Alphabet commenced trading in the Dow Jones Industrial Average on Monday, replacing Verizon after 22 years and becoming the fifth Magnificent Seven component in the 30-stock benchmark; its shares advanced about 4%. CNBC US News The Dow closed above 52,000 for the first time, with technology broadly leading the market to open the holiday-shortened week; the Nasdaq-100 gained over 2%, and Tesla surged more than 8%. TheStreet

X / Social Pulse

  • Anthropic's Saturday statement—"Since June 12, we've been working closely with the US government to restore access to Claude Mythos 5 and Fable 5"—emerged as the weekend's most engaged AI post. Its reception was a mix of relief regarding Mythos 5 and sustained pressure for a definitive Fable 5 restoration timeline in the replies.
  • The controlled rollout of GPT-5.6 dominated developer conversations, with "banned" frequently used rhetorically despite OpenAI's "limited preview" characterization. This sparked a noticeable increase in inquiries into open-weight and local model alternatives, perceived as a hedge against Washington's potential launch interventions.
  • Policy-focused accounts highlighted the deeper anomaly: the classified nature of the "covered frontier model" criteria. This implies models are being subjected to gating protocols before the underlying regulatory standards justifying such restrictions have been publicly disclosed.

One to Watch

Fable 5's restoration. Key indicators for its return include any expansion of the roughly 100-organization Annex A list by Commerce, the July 8 implementation of Anthropic's revised ID-verification privacy policy, and the August 1 executive order deadline for NSA, Treasury, and CISA to finalize the frontier-model review framework. With two leading frontier labs now operating on Washington's timeline, the pace of regulatory approval remains the critical determinant for future roadmaps.

Quick Hits

  • Global AI Statement: Thirty-five nations, including India, Germany, the Netherlands, and the EU, endorsed the Joint Statement on AI Opportunity at the second Pax Silica summit, committing to secure AI, chip, and rare-earth supply chains. GKToday
  • Automation Outlook: Anthropic's June Economic Index found 35% of nearly 9,700 surveyed users anticipate AI handling the majority of their work within 12 months, indicating a market shift from AI augmentation to direct automation in API workflows. Build Fast with AI
  • Efficiency over Expenditure: Enterprises are moving beyond "tokenmaxxing" toward model routing and strict spending caps; Uber reportedly exhausted its annual AI budget in four months and now implements a $1,500 per-employee cap. CNBC
  • Distillation Allegations: Anthropic informed the Senate that Alibaba's Qwen lab executed the largest known distillation attack on Claude, involving 25,000 fake accounts and 28.8 million exchanges from April 22 to June 5. Alibaba denies training on proprietary outputs. Forbes
  • Gemini 3.5 Pro Delay: Google's Gemini 3.5 Pro slipped to a July general availability target, marking its second consecutive missed deadline, following the departure of four senior Gemini researchers within six days this month. Build Fast with AI

The same day Sacramento procured Claude for every agency, Seoul committed half a trillion dollars to the foundational chips. Governments transition from merely regulating the AI race to actively competing within it. With two frontier models gated behind federal review and a third delayed, the bottleneck this summer isn't capability; it's permission.

Sources

California-Anthropic: Governor's office · TechCrunch · CBS Sacramento South Korea: CNN · Al Jazeera · Japan Times Mythos/Fable: CNN · TechTimes · Anthropic on X GPT-5.6: Axios · VentureBeat Markets: CNBC · US News · TheStreet Quick Hits: GKToday · CNBC · Forbes · Build Fast with AI

Lock in. M. mazen@thorterminal.com

Read more