TL;DR
China’s rules for human-like emotional AI services took effect on July 15, 2026. US agencies face an August 1 deadline to create a voluntary frontier-model review system, while another stage of the EU AI Act applies August 2, subject to newly adopted changes delaying high-risk requirements.
China’s new rules for human-like emotional AI services took effect on July 15, 2026, opening a 19-day period that also includes major US and European regulatory deadlines. The clustered dates show governments moving AI scrutiny closer to deployment, although the three systems impose different tests and only China requires a regulatory safety review for the covered services.
China’s five-agency regulation covers services that simulate human personality, thought patterns or communication styles through sustained emotional interaction. Providers must conduct a safety assessment before launching or materially changing a covered service, submit the report to provincial cyberspace authorities and complete applicable algorithm-filing requirements. The rules also require clear AI identity notices, protections for minors and measures addressing dependency and excessive use.
The measure does not cover every chatbot or AI agent. Official guidance excludes customer service, knowledge queries, work assistants, education and research tools when they do not provide sustained emotional interaction. Providers may face warnings, service restrictions or fines for violations, while services involving health or finance must also follow sector-specific rules.
In the United States, Executive Order 14409 gives federal agencies until August 1, 60 days after the order was signed, to develop a classified cyber-capability benchmark and design a voluntary framework with AI developers. Participating companies could give the government access to covered frontier models for up to 30 days before release to trusted partners. The order expressly rejects mandatory federal licensing or preclearance.
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.
AI Prompts for Safety Professionals: Save Hours on Risk Assessments, Incident Reports, Toolbox Talks, and Safety Documentation Using Artificial Intelligence
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Three Systems, Different Tests
The dates reflect a shared government interest in examining some AI systems before broad deployment, but they do not create a single global approval regime. China focuses on content, public interests and risks tied to emotional interaction. The EU uses risk classification, conformity duties and market oversight. The United States is developing a voluntary national-security review tied to cyber capabilities and trusted-partner access.
For companies operating across these markets, one model release may trigger separate reviews of product design, documentation, user protections and security. Compliance work cannot be transferred automatically between jurisdictions because the standards examine different harms and apply to different products.
Regulatory Dates Converge
China’s rules were issued on April 10 by the Cyberspace Administration of China, the National Development and Reform Commission, the Ministry of Industry and Information Technology, the Ministry of Public Security and the State Administration for Market Regulation. They extend China’s existing generative-AI assessment and algorithm-filing system to a defined category of companion-style services.
The EU AI Act has followed a staged schedule. Prohibited practices began applying in February 2025, followed by general-purpose AI and governance provisions in August 2025. Much of the remaining act reaches its application date on August 2, 2026, but that date no longer marks the arrival of every high-risk obligation.
The European Parliament approved the Digital Omnibus changes on June 16, and the Council gave final approval on June 29. The adopted text moves high-risk requirements to December 2, 2027 for stand-alone systems and August 2, 2028 for systems embedded in regulated products. This supersedes the source material’s earlier statement that Council approval was still pending.
“Nothing in the order authorizes a mandatory governmental licensing, preclearance or permitting requirement for releasing new AI models.”
— Executive Order 14409
Scope Gaps Limit Reach
It is not yet clear how many US developers will join the voluntary 30-day framework, how the government will select trusted partners or which models the classified benchmark will identify as covered. Because the benchmark will be classified, outside researchers and customers may have limited visibility into the test.
Enforcement consistency also remains uncertain. China’s provincial reporting process, the EU’s national supervisory structure and the US opt-in system could produce different outcomes for similar models. Open-weight releases are not automatically outside every regime: coverage can depend on where a service is offered, whether a model enters the EU market and whether open-source exemptions or systemic-risk duties apply.
August Deadlines Test Implementation
US agencies are due to establish the classified benchmark and voluntary framework by August 1. On August 2, additional EU AI Act provisions begin applying under the revised schedule, while companies and regulators prepare for later high-risk deadlines. Early enforcement in China will show how authorities interpret emotional interaction, dependency safeguards and the pre-launch safety review.
Key Questions
Did all three jurisdictions create mandatory pre-release approval?
No. China requires a safety assessment for covered anthropomorphic services. The US system is voluntary, while the EU generally relies on risk-based conformity and market rules rather than case-by-case government approval.
Which AI products are covered by China’s new rules?
The rules cover services providing sustained emotional interaction by simulating human personality or communication. Ordinary customer-service bots, work assistants and research tools are excluded when they lack that emotional-interaction function.
What happens under the US framework?
Participating developers may ask the government whether a model meets the covered-frontier threshold and provide secure access for up to 30 days before release to trusted partners.
Does the full EU AI Act take effect on August 2?
Many remaining provisions apply on August 2, 2026, but high-risk requirements have a separate timetable. The adopted Digital Omnibus sets deadlines of December 2027 and August 2028, depending on the system category.
Can one compliance review satisfy all three systems?
Usually not. China examines content and service-specific risks, the EU applies rights and product-safety duties, and the US framework focuses on advanced cyber capabilities and national security.
Source: Thorsten Meyer AI
