AI Policy · Daily

Commerce dropped advanced chip licensing rules for the UAE, drawing a corruption charge from Sen. Warren. Stop AI co-founder Sam Kirchner has been missing since November, when he told members nonviolence may be over; his group warned police he might target OpenAI. Bank of England took direct oversight of AWS, Google Cloud, Oracle and Microsoft as critical third parties under UK financial law, requiring stress tests and mandatory incident reporting. Apple filed a California federal complaint against OpenAI and two former engineers, alleging theft of confidential material. Boko Haram fighters used consumer AI chatbots for weapons support and tactical planning, a Cambridge University field study found.

I.Top Stories

Commerce eases UAE chip export licensing as Warren alleges Trump-family stablecoin tie

The Commerce Department on Friday posted a Federal Register rule that removes license requirements for advanced computing exports to the UAE government and approved firms, including Abu Dhabi AI conglomerate G42 and its cloud subsidiary Core42, and to U.S. companies operating in the country such as Amazon, Apple, xAI, Google, Meta, Microsoft, OpenAI and Oracle. Commerce said it will favorably review license applications involving UAE state investor MGX and cited the UAE's role countering Iran and its position as the largest U.S. trading partner in the Middle East. Sen. Elizabeth Warren, D-Mass., called the provision "corrupt" in a same-day Senate Banking minority statement, tying the eased controls to MGX's use of a stablecoin linked to President Trump's family to fund a $2 billion Binance investment last year. Warren called on Commerce Secretary Howard Lutnick and Under Secretary Jeffrey Kessler to testify before the Senate Banking Committee. The rule is scheduled for official publication July 14.

Read at Financial Times ↗ • Read at CNBC ↗ • Read at Senate Banking ↗

Stop AI co-founder remains missing after telling his group nonviolence may be over

Sam Kirchner, the 27-year-old co-founder of Stop AI, a protest group that demands a permanent global ban on artificial superintelligence, has been missing since Nov. 21, 2025, The Wall Street Journal reported. Four days earlier, he told fellow member Matthew Hall that "the ship may have sailed on nonviolence," and the group called San Francisco police to warn that he might target OpenAI employees; the company locked down its campus, circulated his photo in an internal security warning and told staff to stop wearing company-logo gear in public. Police have not located him, and his fellow members now say he never made a specific threat and that they called authorities out of caution.

Read at WSJ ↗

Bank of England begins regulating AWS, Google Cloud, Oracle and Microsoft as critical financial third parties

Statutory powers took effect Monday designating the four cloud providers as "critical third parties" under UK financial law, per the Guardian. The Bank of England and the Financial Conduct Authority now directly oversee the vendors that underpin UK banking. Named firms must run stress tests and report cyberattacks, outages and natural disaster impacts to both regulators. The Bank of England has separately warned that AI adoption itself could pose financial stability risks as it scales through UK banks.

Read at The Guardian ↗ • Read at Bank of England ↗

Apple sues OpenAI in California federal court over alleged trade secret theft on device plans

Apple filed suit Friday against OpenAI and two former Apple employees in California federal court, alleging OpenAI's senior leadership directed the misappropriation of confidential information about Apple's internal tools, processes and unreleased products. The complaint says former Apple engineers carried confidential material to OpenAI to advance the AI company's hardware ambitions. WSJ characterized the filing as Apple's "thermonuclear" litigation playbook, echoing the Steve Jobs era approach Apple used against Android.

Read at WSJ ↗ • Read at Washington Post ↗ • Read at TechCrunch ↗

Boko Haram fighters used consumer AI chatbots for tactical planning and weapons work in Nigeria

A Cambridge University field study by researcher Antonia Juelich, based on 57 interviews with 27 former Boko Haram members in northeast Nigeria, found that both major factions used consumer AI chatbots across weapons support, tactical planning, operational security and post-mission review, per the New York Times. Former commanders told Juelich the group consulted AI on how to modify motorcycles to jump a defensive trench during an attack on a Nigerian military base. Named tools included ChatGPT, Claude, Gemini, Grok, Meta AI and DeepSeek, per HSToday, and trained users worked around platform safety controls. U.S. military and counterterrorism officials confirmed to the Times that battlefield use is a shift from earlier jihadist AI activity focused on propaganda, translation and recruitment.

Read at NYT ↗

II.China Watch

Beijing sets 2030 targets for a next generation internet built around AI agents

China's Ministry of Industry and Information Technology issued joint guidelines Monday with three sister agencies to overhaul the country's internet basic resources, calling for major advances by 2030 and a more advanced national infrastructure by 2035, TechNode reported. The document directs research into agent-to-agent networks, satellite internet, digital identity infrastructure and IPv6 upgrades, and calls for integrating AI, blockchain and distributed identifiers into the resource system. The new target builds on work covered in AIPD's June 26th edition, when China's State Administration for Market Regulation released seven national standards for AI agent interconnection covering identity, discovery and tool invocation.

Read at TechNode ↗

Ant Group open sources AI agent safety guardrails days after Beijing's Claude Code alert

Ant Group's AI Safety Lab has open sourced two guardrail models, SingGuard-NSFA for autonomous agents and SingGuard for multimodal systems, TechNode reported. SingGuard-NSFA is designed to detect prompt injection, sensitive data theft, malicious code execution, resource abuse and permission misuse before agents take action, and comes in four sizes from 0.8 billion to 9 billion parameters. Ant said the model can make a single risk judgment in about 50 milliseconds and covers seven major risk categories across 133 languages. The release lands five days after China's industry ministry issued a "backdoor" alert on Anthropic's Claude Code.

Read at TechNode ↗

Zhipu AI raises HK$31.4 billion in Hong Kong to fund an AGI research push

Zhipu AI on Monday completed a placement of new H shares raising approximately HK$31.4 billion (about $4 billion), per Jiemian. The company priced 19.78 million new shares at HK$1,588 (about $200) apiece, roughly a 13% discount to the prior close. The raise deepens a summer of megafundraises among Chinese frontier labs; MiniMax's HK$16 billion (about $2 billion) round appeared in AIPD's July 10th edition. In a separate internal letter reported by LatePost, founder Tang Jie said Zhipu will launch a "Touch High" plan focused on artificial general intelligence (AGI) research rather than short-term commercialization, prioritizing long horizon tasks, autonomous agent systems and mechanistic interpretability.

Read at Jiemian ↗

Chinese team claims optical chip design boosts AI inference 100 fold with a fraction of the compute

Chinese researchers have built an all-optical interconnect linking standard electronic chips that they claim boosts AI distributed inference speeds by more than 100 times while using just one-ninth of the typical computational resources, the South China Morning Post reported. Corresponding authors from Peking University published the work in the journal National Science Review. The comparison covers distributed inference on conventional GPU stacks rather than a single accelerator benchmark, and independent evaluations of the reported numbers have not been published outside the paper. Chinese researchers have increasingly pitched optical and photonic architectures as an alternative to piling on more GPUs, a route constrained for Chinese labs by U.S. controls on Nvidia's frontier accelerators.

Read at South China Morning Post ↗

State media warn Chinese bloggers that amateur AI typhoon forecasts may break the law

As Typhoon Bavi bore down on eastern China, state broadcaster China Media Group warned that hobbyist weather bloggers posting AI generated typhoon forecasts on social media could run afoul of the law, the South China Morning Post reported. Some had drawn on open source AI weather models to predict landfall zones for specific locations, and a few were charging fees for tailored outputs. Under China's Meteorology Law, only official meteorological stations may issue public weather forecasts and warnings through a centralized release system. The rebuke arrives as Chinese authorities have already moved to bring AI generated content under labeling and traceability rules, extending existing content controls to a class of open weight scientific models that had largely escaped the framework.

Read at South China Morning Post ↗

III.Policy Tracker

Labor Department opens comment period on nationwide AI usage survey to field January 2027

The Bureau of Labor Statistics opened a two-month public comment period Friday via Federal Register notice on adding an AI usage module to the American Time Use Survey, the federal government's standing time diary study, per FedScoop. The module is scheduled to field for two years beginning January 2027, covering whether respondents use AI tools, what tasks they use them for, how usage varies by demographics and occupation, and how AI use maps to time spent on work, household, education and leisure. BLS is seeking OMB clearance for the module, which would be a new federal statistical instrument linking self-reported AI usage to time diary activity data.

Read at FedScoop ↗

Data centers become 2026 gubernatorial race flashpoint on energy prices and land use

Political backlash to AI data centers has become a first-tier issue in 2026 gubernatorial contests, forcing incumbents and challengers to answer on energy prices, water and land use, The Hill reported. Voter anger over rising electricity bills is looming over midterm races more broadly, per AP, with advocacy group PowerLines reporting that gas and electric utilities sought or secured more than $34 billion in rate increases in the first three quarters of 2025, roughly double the same period a year earlier. In Georgia, a data center hub, Georgia Power has proposed spending $15 billion to expand generating capacity primarily to meet data center demand, with average residential bills already at $175 per month.

Read at AP ↗ • Read at The Hill ↗

IV.Industry & Market Watch

White House steers tech companies to Intel, bending the market

The White House has pressed Intel's potential customers and partners, including Apple, Nvidia and SpaceX, into commercial arrangements with the chipmaker, the Wall Street Journal reported. President Trump and Commerce Secretary Howard Lutnick urged Apple CEO Tim Cook last year to use Intel's manufacturing facilities during discussions of proposed semiconductor tariffs; Apple later secured a tariff exemption after committing to expanded U.S. investments and now plans to have Intel manufacture chips for certain Mac and iPhone products. The administration converted $9 billion in federal grants into a 10% Intel equity stake, making the U.S. government the company's largest shareholder. Intel shares have more than quadrupled since CEO Lip-Bu Tan took over in March 2025.

Read at WSJ ↗ • Read at Investing.com ↗

Meta doubles Louisiana AI data center to 5 gigawatts at $50 billion, campus total set to exceed $250 billion

Meta on Monday said its Richland Parish, Louisiana data center will expand to 5 gigawatts of compute capacity for more than $50 billion in investment, up from the previously projected 2 gigawatts, per Reuters. Total expected investment across the Louisiana campus will surpass $250 billion after Meta added roughly $40 billion, Bloomberg reported. The facility is backed by state tax incentives. Since breaking ground in December 2024, local Louisiana businesses have received more than $1.6 billion in Meta contracts, and Meta plans another $1 billion for local roads, water and wastewater systems. Meta has pledged $600 billion in U.S. infrastructure and jobs over the next three years.

Read at Bloomberg ↗ • Read at Investing.com ↗

DoorDash and Siemens join Airbnb in adopting Chinese AI models to cut costs

DoorDash and Siemens are using Chinese AI models to cut ballooning enterprise bills and reduce reliance on U.S. technology, joining Airbnb in a pattern flagged by House investigators, the Financial Times reported. The House probe of Airbnb and Anysphere's use of Alibaba's Qwen and Moonshot's Kimi models was reported by CNBC in AIPD's July 8th edition, based on April letters from House Homeland Security Chair Andrew Garbarino and China Select Committee Chair John Moolenaar. The FT reporting expands the target set of the House inquiry from a two-company probe to a cross-sector one.

Read at Financial Times ↗

Axios maps an AI class divide of "haves," "have-nots" and "know-nots"

Frontier models such as OpenAI's Sol and Anthropic's Fable are built for the agentic coding and research most Americans will never touch, while millions encounter AI passively through search summaries, customer-service bots and features they don't know are running, Axios wrote Friday. Nearly half of U.S. adults use AI chatbots, but the most common use is basic information search, and the group running agentic tools is a fraction of OpenAI's more than 50 million paying subscribers, who are themselves a slice of its more than 900 million weekly ChatGPT users. Adoption is rising while confidence falls: 63% of Americans say AI is advancing too quickly and 16% expect it to benefit society over the next 20 years, per Pew Research Center data cited in the piece.

Read at Axios ↗

V.Global & Geopolitics

EU Commission preliminarily finds Meta's Instagram and Facebook breach DSA over addictive design

The European Commission on Friday preliminarily concluded that Meta's Instagram and Facebook breach the Digital Services Act by using infinite scroll, autoplay, push notifications and highly personalized recommendations that regulators say shift users into "autopilot mode." The Commission found Meta had failed to assess the risks its platforms pose to users' physical and mental health, including minors and vulnerable adults, and said existing time management tools could be easily dismissed. Executive Vice-President for Tech Sovereignty Henna Virkkunen said the DSA "provides a clear framework to hold platforms accountable for the addictive design and effects of their services." A confirmed finding after Meta's response could carry fines up to 6% of the company's global annual revenue.

Read at AP ↗ • Read at TechCrunch ↗ • Read at European Commission ↗

Von der Leyen proposes EU-wide law setting minimum age for social media access

European Commission President Ursula von der Leyen at a Brussels press conference Monday said the EU will propose legislation setting a legal age for social media access, structured around "phased and gradual access for different age ranges." The proposal accompanies a Commission special panel report on child safety online published the same day. Under the outline, children under 13 would access social media only in the presence of an adult, and children under 3 would have no screen exposure at all, per AP. Von der Leyen, a physician by training, said "childhood won't wait and once it's gone, we can never give it back."

Read at Bloomberg ↗ • Read at AP ↗ • Read at Financial Times ↗

Facewatch to send real time police alerts from UK store facial recognition scans

Facial recognition firm Facewatch will begin sending real time police alerts when its live scan systems in more than 100 UK stores, including Sainsbury's, B&M and Spar, flag someone the system tags as a serious offender, the Guardian reported. Rollout is expected in autumn 2026, with Facewatch reporting an average alert time of four seconds. Civil liberties group Liberty warned the technology has outrun regulation.

Read at The Guardian ↗

South Korea's Lee government weighs 'national dividend' from AI driven tax windfall

President Lee Jae Myung's government plans to deploy a record AI driven semiconductor tax windfall into a growth agenda, Bloomberg reported. The fiscal move follows the IMF's upgraded 2026 South Korea growth forecast to 2.6% on AI hardware exports, reported by Bloomberg in AIPD's July 9th edition. Presidential Chief of Staff for Policy Kim Yong-beom earlier this year proposed distributing excess tax revenue from AI infrastructure gains as a "national dividend," pointing to Norway's sovereign wealth fund as a model, per the Korea Times and DONG-A ILBO.

Read at Bloomberg ↗ • Read at Korea Times ↗ • Read at DONG-A ILBO ↗