AI Policy Roundup | Summer 2026

ETA Intelligence  /  AI Policy Roundup

The AI Rulebook Just Got Real: What Changed This Summer

August 2 is a live compliance date on two continents, Congress has its first serious bipartisan AI framework on the table, and federal agencies are buying agentic AI at scale. Here is what business and technology leaders across the United States need to know right now.

Published August 2, 2026  ·  Enterprise Technology Association

For three years, AI policy has been mostly signal: frameworks, guidance, discussion drafts, and executive orders that pointed at future obligations. That period is ending. Several of the dates that lawyers have been circling since 2024 land this week, and the shape of a national framework is finally visible in legislative text rather than in principle documents.

Below are the seven developments most likely to touch your organization's roadmap, your procurement process, or your risk register over the next two quarters.

Seven Policy Moves Leaders Should Know

State Law Operative August 2, 2026

California's AI Transparency Act becomes operative today

The California AI Transparency Act (SB 942), as amended by AB 853, takes effect August 2, 2026. It applies to "covered providers," meaning developers of publicly accessible generative AI systems with more than one million monthly visitors or users in California. Core duties include offering a free AI content detection tool, applying visible manifest disclosures, and embedding latent provenance disclosures in generated content. The original January 1, 2026 date was pushed back by AB 853, which also layers obligations onto large online platforms and capture device manufacturers on later dates in 2027 and 2028.

Why it matters California is now the first state operating a comprehensive provenance and watermarking mandate. Even if your company is well below the user threshold, your vendors may not be. Content provenance metadata is about to start showing up in the assets your marketing, communications, and product teams handle every day, and your contracts with generative AI licensees should reflect that.

Sources California Legislative Information: SB 942 text  ·  A&O Shearman: California's evolving AI legal landscape

Global Applies August 2, 2026

The EU AI Act's transparency rules bite, even though high-risk duties slipped to 2027

August 2, 2026 is the AI Act's general application date. The Digital Omnibus on AI, agreed politically in May and approved by the European Parliament and Council in June, moved standalone Annex III high-risk obligations to December 2, 2027, and high-risk AI embedded in regulated products to August 2, 2028. What the Omnibus did not move is Article 50. From August 2, providers must ensure people know when they are interacting with an AI system, mark synthetic content in machine-readable form, and disclose deepfakes. The Omnibus also adds a new Article 5 prohibition covering AI-generated non-consensual intimate imagery.

Why it matters A lot of US teams read the high-risk delay as a full reprieve. It is not. If you have EU market exposure through a customer base, a subsidiary, or a product that reaches EU users, your chatbot disclosures, synthetic media labeling, and content marking obligations are live now. The 2027 extension buys time for conformity assessments, not for transparency.

Sources European Commission: Navigating the AI Act  ·  Jones Walker: Why August 2 still matters  ·  Gibson Dunn: Omnibus agreement details

Congress June 4, 2026

The Great American AI Act puts a federal framework in writing

Representatives Jay Obernolte (R-CA) and Lori Trahan (D-MA) released a 269-page bipartisan discussion draft of the Great American Artificial Intelligence Act of 2026, joined by Representatives Scott Franklin (R-FL), Suhas Subramanyam (D-VA), Erin Houchin (R-IN), and Scott Peters (D-CA). The draft is organized into four titles: Frontier AI Governance, Workforce, Cybersecurity, and Research, Development, and International Cooperation. It would impose binding development obligations on "large frontier developers," defined as companies with $500 million or more in annual revenue that have trained a frontier model, and it includes a three-year preemption of state laws governing AI development. The bill has not been formally introduced; the sponsors released it to collect public comment. The co-chairs of the House Commission on AI and the Innovation Economy opposed the draft within hours of its release.

Why it matters This is the first credible attempt at a comprehensive federal AI regime, and it is narrow by design: the binding obligations sit on the largest model developers, not on the businesses deploying AI. If it advances in anything like its current form, the compliance map most companies built state by state over the last eighteen months gets simpler. If it stalls, the patchwork holds. Either way, the workforce and cybersecurity titles are worth reading directly.

Sources Rep. Obernolte: official release  ·  Tech Policy Press: full analysis  ·  Roll Call: preemption provisions

Congress June 25, 2026

House Science Committee advances ten AI bills with near-unanimous support

The House Science, Space, and Technology Committee marked up and favorably reported ten AI-related bills in a single session, most of them unanimously. Among them: H.R. 2385, the CREATE AI Act, which establishes the National Artificial Intelligence Research Resource to expand access to compute and datasets for AI research through the National Science Foundation, and H.R. 9363, the AI Security and Innovation Act, which strengthens federal coordination on securing AI systems. The package spans research infrastructure, security, workforce development, transparency, data governance, and data center energy measurement, with NIST and NSF assigned central roles.

Why it matters These bills do not impose prescriptive requirements on private-sector deployers, which is exactly why they moved. What they do is decide who writes the standards. NIST and NSF are being positioned as the technical authority behind whatever federal framework eventually lands, and the research and workforce funding streams they control are ones regional ecosystems, universities, and employers can actually access.

Sources Mintz: AI Washington Report, July 2026

Executive Action June 2, 2026

A new executive order adds cybersecurity review before frontier model releases

President Trump signed an executive order titled "Promoting Advanced Artificial Intelligence Innovation and Security," establishing cybersecurity requirements and a voluntary framework for secure deployment of frontier models. In practice it asks developers to present their most capable models for federal review roughly 30 days before public release. The effect was visible within weeks: OpenAI's GPT-5.6 family went first to a small group of government-approved partners in late June before general availability on July 9. The June order sits alongside the December 2025 order, "Ensuring a National Policy Framework for Artificial Intelligence," which created a Department of Justice AI Litigation Task Force to challenge state AI laws and directed Commerce, FTC, and FCC action on preemption.

Why it matters Two things follow for enterprises. First, model availability timelines are now partly a function of federal review, which affects vendor roadmap planning. Second, the administration's preemption push does not suspend any state law on its own. Until a statute passes or a court rules, California, Texas, Illinois, Utah, and Colorado obligations remain enforceable, and the prudent posture is to keep complying with them.

Sources Lathrop GPM: analysis of the June order  ·  Engadget: how the review affected GPT-5.6's rollout  ·  White & Case: state law compliance takeaways

Procurement July 28, 2026

GSA signs an agentic AI OneGov deal, with governance baked into the pitch

The General Services Administration announced a OneGov offering with CORAS, giving federal agencies discounted access to GARY, an agentic AI orchestrator running on a FedRAMP High, IL 5/6 platform. GSA said agencies get 80 percent off GARY Insight and 40 percent off GARY Execute through September 30, 2027, with an optional 90-day pilot. The stated pitch is a governed digital workforce operating under human-authored rules, with a person approving every result and a full audit trail. GSA also highlighted an "App Factory" capability for building mission-specific applications in days rather than the six to twelve months typical of traditional development.

Why it matters Federal procurement language becomes de facto private-sector standard faster than most people expect. The terms GSA chose to foreground here are worth copying: human approval on every result, human-authored rules, and a complete audit trail. If you are writing an internal agent policy this quarter, that is a defensible baseline and one your auditors will recognize.

Sources GSA: official announcement  ·  Nextgov/FCW: deal terms  ·  FedScoop: reporting

States Effective dates through 2027

The state calendar keeps shifting, and Colorado moved again

Colorado's SB 26-189, signed May 14, 2026, repealed and reenacted the state's AI Act framework. Core automated decision-making duties now begin January 1, 2027 instead of June 30, 2026. Elsewhere: the New York Department of Financial Services published binding AI model risk management guidance for insurers on July 1, 2026, with compliance required by January 1, 2027. Illinois amendments to the AI Video Interview Act took effect July 1, extending the law to AI-generated avatar interviewers. Texas HB 1709, requiring employers using AI in hiring to notify candidates and offer a human review option, passed the Texas House on July 8. California, Texas, Illinois, and Utah rules are already in force.

Why it matters Two practical consequences. First, if your compliance calendar still shows Colorado as a mid-2026 obligation, it is stale; rebuild it around January 2027 and use the extra runway to complete your automated-decision inventory. Second, employment and hiring is where state AI law is converging fastest. Notice requirements and a human review path for candidates are becoming the common denominator across jurisdictions, and building them once is cheaper than building them five times.

Sources Vorp Labs: US AI regulation tracker  ·  Transparency Coalition: legislative update

What to Put on the Calendar

• Now: Confirm whether any vendor in your generative AI stack is a California "covered provider," and check what provenance metadata their outputs now carry.
• Now: Audit customer-facing AI interactions for disclosure. If EU users can reach it, Article 50 applies today.
• This quarter: Complete an automated decision-making inventory ahead of Colorado's January 1, 2027 start. It is the same inventory most other state frameworks will ask for.
• This quarter: Write your agent governance policy around decision rights, human override thresholds, and audit trails. Federal procurement language gives you a defensible template.
• Watch: Whether the Great American AI Act is formally introduced, and whether its three-year preemption survives committee.

This roundup is provided for general information and is not legal advice. Effective dates and obligations shift quickly; confirm applicability with qualified counsel before adjusting policy or contracts.

Do not track this alone.

ETA members get this intelligence early, in context, and alongside the peers working through the same decisions. Join as a Founding Member at the PRO or LEADER level to unlock ETA events, insights, and member benefits nationwide.

PRO

Built for business and technology professionals. Includes the ETA Events Passport, national network access, AI training and certifications, and access to the ETA partner network.

LEADER

Built for executives and senior leaders with teams, budgets, and technology decisions to make. Includes the VIP Events Passport plus ETA leadership programming and events.

Become a Founding Member Compare Membership Levels

Questions about which level fits your team? Reach us at hello@joineta.org.