Tomorrow two AI deadlines hit. Nobody is talking about them.
On June 2, President Trump signed an executive order on AI innovation and security. Most coverage focused on the 30-day pre-brief framework for frontier models. But the EO contains two other 30-day deadlines that expire tomorrow — July 2 — and a 60-day deadline on August 1 that will reshape how every AI company in America goes to market. The clearinghouse that gets launched tomorrow will be the first permanent government body with authority over AI vulnerability information. The framework that goes live August 1 will determine whether the Fable 5 shutdown was a one-time event or the beginning of a new normal. This is the most important week in AI regulation since the Fable 5 ban — and it is happening almost entirely out of public view.
THE STORY
Executive Order 14409, signed June 2, 2026, contains three deadlines. One of them — the 30-day pre-brief framework — became famous because Fable 5 launched before it was implemented and the government used export controls to enforce compliance. The other two deadlines are arriving tomorrow.
The first deadline: the AI Cybersecurity Clearinghouse. By July 2, the Secretary of the Treasury, in consultation with the National Security Agency and CISA, must form an "AI cybersecurity clearinghouse" in voluntary collaboration with AI industry and critical infrastructure operators. The clearinghouse's mandate: coordinate scanning for software vulnerabilities across AI systems, validate discovered vulnerabilities, and coordinate and prioritize remediation and patch distribution.
This is not a press release. It is a standing infrastructure body with real operational authority. The clearinghouse will have the ability to aggregate vulnerability information from AI labs, government agencies, and critical infrastructure operators — and coordinate how that information is shared and acted upon. Companies that participate early will gain access to coordinated vulnerability intelligence before it is widely known. Companies that do not may find themselves receiving warnings about vulnerabilities in their AI systems only after a coordinated disclosure process they had no role in shaping.
Companies developing or deploying frontier AI systems should consider engaging proactively with the agencies responsible for designing the voluntary framework to position themselves for trusted-partner designation. The clearinghouse's governance structure, data-sharing protocols, and liability protections for participants remain undefined as of this writing — which means the companies that engage earliest will shape the norms that everyone else follows.
The second deadline: CISA Binding Operational Directives. Also by July 2, CISA must release Binding Operational Directives and other guidance to expedite the cyber defense of civilian federal systems — including explicitly requiring the establishment or expansion of AI-enabled defensive tools across the federal government. These BODs carry real compliance weight for every agency and contractor operating civilian federal systems. The directive also facilitates access to cybersecurity tools for state and local authorities and critical infrastructure operators including rural hospitals, community banks, and local utilities.
The practical effect: every hospital, bank, and utility that receives federal funding for cybersecurity will be directed toward specific AI-enabled defensive tools. The companies whose tools are referenced in those BODs gain federal distribution at scale.
The third deadline: August 1. The EO's 60-day deadline — August 1 — is when a multi-agency group led by Treasury, NSA, and CISA must finalize the voluntary pre-release framework that defines "covered frontier models" and establishes how developers engage with the government before releasing them to trusted partners. This is the framework that, if it had existed on June 2, would have given Anthropic a formal process to follow before launching Fable 5. Its absence on June 9 is what made the June 12 export control action possible.
The voluntary pre-release framework for pre-release government engagement is due to be finalized by August 1, 2026. While the Order generally contemplates a framework that would guide developers on how to provide the federal government with access to covered frontier models for up to 30 days before release to "trusted partners," the specific mechanics, scope, and conditions of that framework have not yet been designed.
Once the framework is finalized, every frontier AI lab launching a new model will have a formal procedure to follow — or will be taking the same regulatory risk Anthropic took on June 9. The August 1 deadline is the most important date in AI regulation for the rest of 2026. It is also the date from which the next Fable 5 scenario becomes either impossible or predictable.
Meanwhile, the week that just ended produced its own summary. The Magnificent Seven plus Broadcom and Oracle erased $2.7 trillion in market value in June — the first significant repricing of AI stocks on ROI concerns rather than macro events. GPT-5.6 launched in limited government-gated preview. Mythos 5 returned for approximately 100 approved US organizations. Gemini 3.5 Pro missed its June 30 deadline. OpenAI unveiled the Jalapeño chip with Broadcom. Noam Shazeer — the Transformer's co-inventor — joined OpenAI. Anthropic accused Alibaba of 28.8 million distillation attacks. China announced a $295 billion five-year AI infrastructure plan.
June 2026 was the month AI grew up. July is when it has to prove it.
THE MONEY ANGLE
1. The AI Cybersecurity Clearinghouse is a new market — and it opens tomorrow. The Treasury-led clearinghouse creates a permanent public-private body for AI vulnerability intelligence. Companies that participate gain early access to government-identified vulnerabilities, validated threat assessments, and coordinated patch distribution. The clearinghouse also positions participants to shape its governance norms during the formative period. The market opportunity: every AI security company, every critical infrastructure operator, and every enterprise with federal contracts now has a new regulatory body to engage with. The vendors who build the tools the clearinghouse will use — vulnerability scanners, AI-enabled threat detection, patch distribution infrastructure — are the quietest beneficiaries of June 2's executive order.
2. The August 1 framework is the IPO document nobody is reading yet. Both OpenAI and Anthropic have filed confidential S-1s targeting IPOs in the second half of 2026. Both will need to describe in their risk factors how the August 1 voluntary pre-release framework applies to their model deployment practices. The company that has already agreed to the framework — and is therefore positioned as a cooperative government partner — will describe a materially lower regulatory risk than the company that has not. Anthropic's negotiation over Fable 5 is the most concrete available signal of where that negotiation stands. OpenAI's relative silence on the framework suggests it has either already agreed or is deliberately staying quiet ahead of the roadshow. Either way, the August 1 disclosure will be the most important risk factor section in either prospectus.
3. The week's $2.7 trillion reset created the most attractive AI infrastructure entry point of the year. The companies that lost the most market value in June — Nvidia, Broadcom, Microsoft, Meta, Amazon — are the same companies whose AI infrastructure buildout is supported by the new regulatory framework and government demand for AI-enabled cybersecurity tools. The CISA BODs that go live tomorrow will direct federal spending toward AI-enabled defensive tools. The clearinghouse creates government demand for vulnerability scanning infrastructure. The framework creates a trusted-partner designation process that rewards the companies already most embedded in government AI contracts. The $2.7 trillion in erased value is partially the speculative premium coming out. The infrastructure build is still happening.
THE OPPORTUNITY
1. July 2 is the day to engage with the AI Cybersecurity Clearinghouse. The clearinghouse's governance structure, data-sharing protocols, and liability protections remain undefined. That means the first participants will set the norms. If your company operates critical infrastructure, develops AI-enabled security tools, or has federal contracts involving AI systems — contact the Treasury Department's cybersecurity office this week. The companies that engage in the first 30 days shape the clearinghouse's practices. The companies that engage in month six follow the practices others set. Business owner move | Founder move
2. Map your model release calendar against the August 1 framework. If you're an AI developer with a model launch planned for Q3 or Q4 2026, the August 1 framework is the compliance clock that matters. The framework is voluntary — but as the Fable 5 case demonstrated, the government has export control authority to enforce the spirit of the framework even before its letter is finalized. The safe assumption: any model launch after August 1 should budget 30 days for voluntary government pre-brief, even if formal requirements haven't been codified. The cost of the pre-brief is operational delay. The cost of skipping it — as Anthropic learned — is a global shutdown. Business owner move | Founder move
3. The CISA BODs create immediate procurement opportunities. CISA's Binding Operational Directives, due tomorrow, will specify AI-enabled defensive tools that federal agencies, state and local authorities, and critical infrastructure operators must adopt. The vendors referenced in those BODs gain federal distribution at scale — without a competitive procurement process. If you sell AI security, vulnerability management, or cyber defense tools, obtain the BODs tomorrow and assess whether your products qualify for reference. The distribution opportunity in the next 90 days is larger than any enterprise sales motion you're likely to run this quarter. Business owner move | Founder move
THE WEEK IN NUMBERS
The June 2026 scorecard — everything that mattered:
$2.7T — market cap erased, AI sector, June $295B — China five-year AI infrastructure plan $60B — SpaceX acquisition of Cursor $6.3B — Reflection AI compute deal with SpaceX (Colossus) $4.8B — estimated annual Colossus external compute revenue 9 months — Jalapeño design cycle, OpenAI × Broadcom 200% — Getty Images stock move on OpenAI deal 16 days — Fable 5 offline as of today ~100 — organizations approved for Mythos 5 (Lutnick letter) 8% — Manifold odds of full Fable 5 restoration before June 30 4 — senior Google DeepMind researchers departed in six days 28.8M — fraudulent exchanges Anthropic attributed to Alibaba $965B — Anthropic IPO target valuation $850B–$1T — OpenAI IPO target valuation July 2 — tomorrow: AI Clearinghouse + CISA BODs due August 1 — frontier model framework finalized
QUICK HITS
Gemini 3.5 Pro Misses June 30 — Again Google's flagship next-generation model did not launch before June 30 despite Sundar Pichai's explicit commitment at Google I/O on May 19. The model reportedly underwhelmed in enterprise testing. Four senior DeepMind researchers departed for competitors in the same week. Google has not announced a revised launch date. Prediction markets moved from 50–55% odds of a June launch to near-zero after June 28 without an announcement. ◆ Money angle: A missed committed launch date, combined with talent attrition and a $400 billion market cap decline in June, creates a credibility problem for Google's AI roadmap. The Gemini 3.5 Pro delay is the most visible indicator of a company under competitive pressure across all three dimensions simultaneously: product, talent, and market value.
Anthropic Files Identity Verification System for Fable 5 Anthropic updated its privacy policy, effective July 8, 2026, to collect government-issued ID and biometrics for Fable 5 access. The identity verification system — built on Peter Thiel-backed Persona — is likely the technical mechanism for restoring Fable 5 to US users while maintaining compliance with the export control directive banning access by foreign nationals. If the system works as described, US passport or driver's license holders could regain access within weeks. ◆ Money angle: Identity verification at the model access layer is a $3–5 billion market that Forrester called "the era of KYC for AI" two weeks ago. Anthropic just became the first frontier AI company to implement it in production. Every other frontier lab now has a reference implementation to evaluate — and a compliance standard to match before their next major model launch.
The Week's Biggest Losers and Winners Losers: Nvidia (-12% in June), Meta (-9%), Microsoft (-8%), Alphabet (-11%). Winners: Broadcom (up on Jalapeño partnership announcement), CoreWeave (private, but term sheet activity accelerated), Anthropic (private, but Mythos partial restoration and identity verification progress signals IPO timeline on track), Jane Street (disclosed $1B CoreWeave stake performing well). The common thread in the winners: infrastructure with government relationships. The common thread in the losers: application companies whose AI revenue has not yet justified their capex commitments. ◆ Money angle: The June reset separated the infrastructure layer from the application layer for the first time in AI investing. That separation is likely permanent until the application companies demonstrate earnings multiples that justify their current prices.
June ended with more regulatory infrastructure, more government control over frontier AI deployment, and less speculative premium in AI stocks than it started with. July opens with two deadlines, an identity verification system, and the beginning of the most consequential AI regulatory framework ever written. The future is still profitable. It is just more complicated than it looked on January 1.
See you tomorrow, The Future Geek Team