TL;DR
China, the United States and the European Union are activating or completing separate AI pre-release regimes between July 15 and August 2, 2026. The systems pursue different goals, and key questions remain about US participation, possible EU deadline changes and coverage of foreign open-weight models.
China, the United States and the European Union are activating or completing separate systems for reviewing advanced artificial intelligence within 19 days, creating a concentrated compliance period for companies operating across the three markets. China’s rules took effect July 15, while US and EU milestones are scheduled for August 1 and August 2, according to the supplied Thorsten Meyer AI dispatch.
China’s Interim Measures for AI Anthropomorphic Interaction Services extend the country’s existing approval approach to human-like systems, including AI companions and agents, the dispatch reports. The measures were issued in April by five agencies, including the Cyberspace Administration of China, the national development and industry ministries, public security authorities and the market regulator.
Under the Chinese model described by the dispatch, covered services can face security review before public deployment, algorithm registration and regulator-directed design changes. Providers also face continuing duties, including a reported 24-hour incident deadline, responses to government information requests within 48 hours and implementation of ordered algorithm changes.
The US framework scheduled for August 1 is different. The dispatch says Executive Order 14409 establishes a voluntary arrangement under which participating frontier-model developers provide the government with access for 30 days before release. Evaluations use classified criteria, while trusted-partner status and federal procurement access are presented as incentives. On August 2, the EU AI Act reaches its broad application milestone, bringing risk classification, conformity reviews, technical records and post-market monitoring into operation across covered systems.
Three Gates Close in Nineteen Days
The Pre-Release Regime Goes Global
Same-day-verified · one instinct, three architectures — and none of them binds the open frontier
Anthropomorphic-interaction measures take effect: five agencies extend the CAC approval regime to companion AI and agents.
EO 14409’s classified benchmark and voluntary 30-day pre-release framework harden. NSA designates covered frontier models.
The AI Act becomes fully applicable — the staged rollout that began February 2025 reaches its final station.
Same instinct, three theories of a gate
STEELMAN: THE GATE-SKEPTIC CASE
Pre-release regimes structurally favor incumbents who can afford the process — and none of the three binds an open-weight release from a lab outside its jurisdiction. The gates go up exactly as the fastest-moving part of the frontier walks around them.
The signal: a model can clear all three gates having been evaluated for three almost non-overlapping things — content control, fundamental rights, national security. Jurisdiction is now an architectural property. If your deployment calendar doesn’t carry July 15, August 1, and August 2, it’s a calendar for a market you’re not in.
As an affiliate, we earn on qualifying purchases.
Global Launch Calendars Face New Gates
The clustered dates mean developers serving several markets can no longer treat compliance as a review conducted after a product is released. Deployment timing, model access and technical documentation may now need to be planned around three state systems with different tests and enforcement tools.
The overlap does not amount to a shared international standard. China’s framework centers on content control and social stability; the EU system emphasizes fundamental rights and product safety; and the US arrangement concentrates on national-security evaluation. A model could pass through all three systems without being examined against a common set of risks.
Three Regulatory Models Take Shape
China has required security reviews and algorithm filings for covered generative AI services since 2023. The July measures apply that established regulatory pattern to anthropomorphic AI interactions, giving authorities a direct role in the design and operation of companions and autonomous agents.
The EU AI Act has entered force in stages. Prohibited practices began applying in February 2025, followed by obligations for general-purpose AI models in August 2025. The August 2026 milestone activates most remaining provisions, including controls for high-risk systems. The United Kingdom remains outside this three-part sequence, relying on sector regulators and principles-based rules rather than a formal central approval gate.
“Jurisdiction is now an architectural property.”
— Thorsten Meyer AI dispatch
Deadline Changes and Coverage Gaps
The EU schedule could still change for some high-risk systems. The dispatch says a Digital Omnibus proposal, approved by the European Parliament on June 16 by 423 votes to 57, with 174 abstentions, would move certain deadlines. It has not yet completed Council adoption or publication in the Official Journal, so the dispatch treats August 2 as the operative date. The exact provisions affected by any final measure require confirmation from the adopted legal text.
Questions also remain about the US program’s participation rate, classified benchmarks and enforcement reach. Because the framework is described as voluntary, it is not yet clear how many developers will provide early access or how federal procurement incentives will influence participation. None of the three systems appears to fully bind an open-weight model released abroad by a developer outside the relevant jurisdiction, leaving a possible coverage gap.
US and EU Milestones Arrive
Attention now moves to August 1, when the US evaluation framework is scheduled to harden and the National Security Agency is expected, according to the dispatch, to designate covered frontier models. Agencies and participating developers may then provide more information about eligibility, access procedures and trusted-partner benefits.
One day later, companies operating in Europe must work from the EU AI Act requirements legally applicable on August 2, unless pending legislation changes specific dates before then. Regulators’ implementation guidance, enforcement decisions and treatment of cross-border or open-weight releases will show how far the three gates reach in practice.
Key Questions
What changed on July 15?
China’s anthropomorphic-interaction measures took effect, extending its regulatory system to covered human-like AI companions and agents, according to the dispatch.
Does the United States require approval before an AI model is released?
Not under the framework described in the source. The US program offers a voluntary 30-day evaluation window supported by procurement and trusted-partner incentives, rather than mandatory approval for every model.
Is the entire EU AI Act first taking effect on August 2?
No. Several provisions already apply, including prohibited-practice rules from February 2025 and general-purpose AI obligations from August 2025. August 2, 2026 is the broad application milestone for most remaining provisions.
Have the EU deadlines already been postponed?
Not on the account provided by the dispatch. A proposal would move some high-risk AI deadlines, but it still requires completion of the legislative process and Official Journal publication before changing the operative schedule.
Do the three regimes apply the same safety test?
No. They address different policy objectives: Chinese state controls, European rights and product-safety requirements, and US national-security evaluation. Passing one process does not establish compliance with the other two systems.
Source: Thorsten Meyer AI
