AI Regulation Tracker

China makes digital assets, including game inventories, legally inheritable.

a court ordered a game company to transfer 87 accounts to a deceased player's mother; virtual property has been inheritable under Civil Code Article 127 since 2021, so this is the code being applied, not a new law

China's first law for human-like AI companions comes into force; Doubao deletes its custom agents.

Doubao deleted its custom agents in response

South Korea opens its first criminal case over AI smart glasses used in a state licensing exam.

The FCC adds advanced robotic devices to its Covered List, blocking equipment authorization for new foreign-made home robots.

EU AI Act Article 50 transparency rules apply: chatbots must disclose they are AI, deepfakes must be labeled.

AI Act Article 50: chatbots must disclose they are AI, deepfakes must be labeled

OpenAI pays $3.2 million to settle DOJ claims it discouraged US applicants in hiring.

House Democrats propose a per-token tax on AI companies to fund programs for displaced workers.

Seoul court upholds an 18-month sentence for leaking SK Hynix trade secrets to a Chinese firm.

More than 530 US counties and municipalities now restrict or block new data centers.

more than 530 jurisdictions restrict or block new data centers

Meta's smart glasses are banned from every court in England and Wales.

Estonia plans state-issued ID codes for AI agents with scoped, auditable permissions.

Connecticut judge sanctions a plaintiff for hiding AI prompts in court filings.

A Florida officer is charged with official misconduct after running 717 Flock license-plate searches, including on his wife's car.

OpenAI launches a separate ChatGPT for teens that asks guiding questions instead of giving answers.

a product launch, not a law, ban, ruling or enforcement action

Apple Music tells partners a Made With AI label will mark any track materially generated with AI; over a third of monthly uploads are fully AI-generated.

Nevada approves up to 5,000 paid driverless Tesla vehicles in Clark County over 12 months, a month after allowing ten.

Texas requires local community approval for new data centers and audits power and water use.

Chinese courts rule that replacing a worker with AI is not valid grounds for dismissal, upholding an illegal-dismissal finding in Hangzhou.

Meta settles child-safety claims by 29 US states and territories for $18 billion over 10 years and agrees to a two-hour daily teen limit across Facebook and Instagram.

$18 billion over 10 years plus a two-hour daily cap for teens

Bill Gates publishes a 6,000-word essay proposing that AI be taxed like employees to fund the transition.

an essay proposing a policy, not a policy

Australia's Greens introduce a Senate bill for a 12-month freeze on importing smart glasses and wearable recording devices after $89 Kmart camera glasses sold out.

A federal judge rules the Pentagon's supply-chain-risk designation of Anthropic unlawful retaliation and orders it removed.

Bernie Sanders announces a bill to ban Flock Safety's AI license plate cameras, which he says number more than 120,000 across 6,000 US communities.

Sony Music Publishing and Warner Chappell sue Anthropic and its founders, seeking up to $150,000 per allegedly infringed song.

seeking up to $150,000 per allegedly infringed work

ICE plans to buy Boston Dynamics Spot robot dogs with a budget of $1 to 2 million, alongside a separate $16.7 million DHS surveillance contract.

The Verge reports a national movement of residents destroying Flock cameras: 162 destroyed per FlockStats, and more than 20 cities dropped Flock contracts in July.

162 cameras destroyed per FlockStats; reporting on a movement, not an act of state

ARIA bars wholly AI-generated songs from Australia's official charts and the ARIA Awards; tracks must be substantially human made.

Apple accuses OpenAI of destroying evidence in its trade-secrets case; the filing says an ex-engineer's AI agent ran simulations on a stolen chip schematic. Hearing October 1.

The Pentagon adds ChatGPT Mil and Grok for Government to GenAI.mil for about 3 million personnel; Claude stays off the platform after the guardrail dispute.

ChatGPT Mil and Grok for Government on GenAI.mil for about 3 million personnel; Claude stays off

Tesla's NHTSA crash report lists driver assist as engaged at 104 mph in a fatal May crash in Clute, Texas that police attributed to a medical episode; key fields redacted.

The US Justice Department backs OpenAI's fair use defense against The New York Times.

Norway's digital minister says the government is considering rules for camera-enabled glasses.

Kahn v. Anthropic, a class action over the Claude Max 5x and 20x plans, resurfaces; the complaint says weekly caps make the $200 plan far less than 20 times Pro.

New York City bans student-facing generative AI through 8th grade for 600,000 students.

student-facing generative AI barred through 8th grade for 600,000 students

OpenAI tells House Democrats it is building automated shutdown capabilities after the Hugging Face incident.

Sen. Bernie Sanders and Rep. Greg Casar announce the Ban Artificial Superintelligence Act: a permanent ban on superintelligent AI with penalties up to 20 years in prison and a corporate death penalty.

Seattle Times and Newsday sue OpenAI and Microsoft in SDNY, asking for impoundment and/or destruction of datasets and models trained on their articles.

NHTSA opens an investigation into how Tesla self-certified the Cybercab, which has no steering wheel, pedals or mirrors, as meeting federal safety standards.

OpenAI admits it did not disclose the German wiki hijack by its agents, classing it as misalignment rather than a security incident, and promises a disclosure framework.

California opens an investigation into OpenAI over the July Hugging Face intrusion by its agents.

Sen. Josh Hawley opens an inquiry into OpenAI's handling of the Hugging Face breach in a letter to Sam Altman.

Anthropic CEO Dario Amodei calls for pacing frontier AI and commits Anthropic to embedded third-party evaluators with employee-like access; Sam Altman says OpenAI will do the same.

President Trump rejects the labs' AI slowdown call ("whoever wins AI, wins"); PCAST co-chair David Sacks tells OpenAI and Anthropic to pace themselves without demanding regulation in return.

Trump says the only guardrail AI needs is a "STRONG AND SMART" president and rejects calls to slow AI.