Trump landed in Beijing with Nvidia CEO Jensen Huang aboard Air Force One for a summit with Xi Jinping where AI sits alongside trade and the Strait of Hormuz on the agenda. The House Financial Services Committee marks up two AI bills today: the Artificial Intelligence Practices, Logistics, Actions, and Necessities (AI PLAN) Act (H.R. 2152) and the Unleashing AI Innovation in Financial Services Act (H.R. 4801). Anthropic will brief the House Homeland Security Committee behind closed doors on its Mythos cybersecurity model, covering capabilities and national security implications of the system restricted to defensive work under Project Glasswing. OpenAI and Anthropic are opening Washington offices and retaining lobbyists at levels exceeding any prior cycle, according to a New York Times feature on the industry's federal influence push.
I.AI Policy Today
President Trump departed the White House Tuesday afternoon en route to Beijing for the May 14-15 summit with Chinese President Xi Jinping, with AI named one of three top agenda items alongside trade and the Strait of Hormuz, per The Hill. Trump invited Huang by phone as a last-minute addition to a 16-CEO business delegation that also includes Tesla's Elon Musk, Apple's Tim Cook, Goldman Sachs's David Solomon, BlackRock's Larry Fink and Boeing's Kelly Ortberg, AP reported. A senior U.S. official told Yahoo News that the AI talks aim to establish "channels of de-confliction" amid concerns about advanced AI models in cyberespionage, with a planned U.S.-China board of trade and board of investment to be announced. Huang has publicly described the China market as a $50 billion opportunity for Nvidia, and the Trump administration in January approved limited sales of the H200 AI chip to China but excluded the more advanced Blackwell and Rubin chips.
Read at The Hill ↗ • Read at AP ↗ • Read at Yahoo ↗
The House Financial Services Committee holds a full-committee markup at 10 a.m. ET in 2128 Rayburn covering H.R. 2152, the Artificial Intelligence Practices, Logistics, Actions, and Necessities (AI PLAN) Act, and H.R. 4801, the Unleashing AI Innovation in Financial Services Act, per the committee notice. Members will also consider H.R. 2978 (the Guarding Unprotected Aging Retirees from Deception, or GUARD Act), H.R. 5396 (the Price Stability Act of 2025), H.R. 8278 (the Fostering the Use of Technology to Uphold Regulatory Effectiveness in Supervision, or FUTURES Act) and H.R. 8671 (the Bank Fraud Technology Advancement Act of 2026).
Read at House Financial Services ↗
Anthropic will brief the House Committee on Homeland Security today in a closed-door meeting on its Mythos cybersecurity model, The Hill reported, citing a source familiar with the meeting. The session will focus on the "capabilities, national security implications" of the model. Anthropic has limited Mythos to defensive cybersecurity work through Project Glasswing, a partnership including Amazon Web Services, Apple, Google, Microsoft and Nvidia, per UPI.
Read at The Hill ↗ • Read at UPI ↗
OpenAI will host the grand opening of its first Washington lobbying office located blocks from the White House on Wednesday, having doubled its federal lobbying spend to $1 million in Q1 from a year earlier, the New York Times reports. Anthropic, which opened its own Washington office in April amid its Pentagon supply-chain risk dispute, hired six lobbying firms in recent months and increased its lobbying spend tenfold to $3 million last year, naming Anthony Cimino as its first head of lobbying in January. A Public Citizen analysis found that a quarter of the 13,000 federal lobbyists in Washington now work on AI issues, up from 11 percent in 2023, while Meta, Nvidia and Alphabet spent a combined $47.8 million on federal lobbying last year, up 22 percent from 2024. The surge comes as the Trump administration weighs new federal oversight of AI models and states introduce dozens of AI bills, with a recent NBC News poll finding 57 percent of registered voters say AI's risks outweigh its benefits.
Read at NYT ↗
Rep. Jim Himes (D-Conn.), the top Democrat on the House Intelligence Committee, said at a Politico Security Summit panel Tuesday that it would be "insane" for U.S. intelligence agencies to lack early access to advanced AI models with hacking and cyberdefense applications, Defense One reported. Himes pointed to the National Security Agency's testing of Mythos, the Anthropic model held back from full public release, as the kind of access he meant. He said the Commerce Department should also have a role in AI policy. He warned that prolonged Defense Department conflict with Anthropic, currently in litigation over its supply chain risk designation, could "massively" set back the intelligence community. The comments come as the Trump administration weighs whether the Commerce Department or the intelligence community should oversee AI model evaluations, as reported by the Washington Post in AIPD's May 12 edition.
Read at Defense One ↗
Rep. Ro Khanna of California, whose congressional district covers much of Silicon Valley, is pitching a presidential bid built on a wealth tax and tougher rhetoric about Big Tech and the "Epstein class," Bloomberg reported in a profile titled "Even Silicon Valley's Congressman Wants to Rein in AI." Khanna spent years cheering on the tech industry and building ties with wealthy donors before distancing himself in late December via social media posts critiquing tech and political figures.
Read at Bloomberg ↗
Andreessen Horowitz and its co-founders Marc Andreessen and Ben Horowitz have funneled more than $115.5 million in disclosed federal contributions into the 2026 midterm cycle, making the venture firm the largest known donor of the cycle ahead of George Soros ($102.9 million) and Elon Musk ($85 million), per a New York Times analysis. The firm has put $47.5 million into the crypto super PAC network Fairshake since Election Day 2024 and $50 million into Leading the Future, a pro-AI super PAC modeled on Fairshake that it helped found, alongside $12 million to Trump's MAGA Inc. super PAC. Andreessen Horowitz has also seeded a new AI-focused advocacy nonprofit, American Innovators Network, whose donations are not required to be disclosed, while a competing AI safety super PAC, Public First, was launched in reaction to the firm's spending and has backed candidates including New York state legislator Alex Bores. The firm's $115 million haul dwarfs the roughly $63 million it spent in the 2024 cycle and the $2 million it gave in 2022, and far outpaces venture rivals like Sequoia Capital and Founders Fund, which the Times reports have done nothing comparable.
Read at NYT ↗
The Senate Judiciary Subcommittee on Privacy, Technology, and the Law convenes at 2:30 p.m. ET in 226 Dirksen to examine how social media verdicts demand federal action to protect kids online, per the committee notice. The hearing comes as families have brought wrongful death suits against OpenAI seeking to apply consumer product safety standards to chatbot makers, as reported by the New York Times in AIPD's May 12 edition.
Read at Senate Judiciary ↗
II.China Watch
On Alibaba's FY2026 Q4 analyst call, Chairman and Chief Executive Wu Yongming said annualized recurring revenue from AI models and application services, including the Bailian model-as-a-service platform, will exceed 10 billion yuan (about $1.4 billion) in the June quarter and surpass 30 billion yuan (about $4.1 billion) by year-end, per Jiemian. Wu said the segment is a high margin pillar of future revenue and characterized the growth trajectory as exponential. The figures give a public benchmark for Chinese AI model commercialization as Beijing pushes domestic adoption under U.S. export controls on advanced accelerators.
Read at Jiemian ↗
OpenBMB and Tsinghua University released MiniCPM-V 4.6 under the Apache 2.0 license as a 1.3 billion parameter vision language model that matches larger competitors on key benchmarks, per Pandaily. The model is designed to run on a single Nvidia RTX 4090, putting capable vision language inference within reach of consumer hardware. Apache 2.0 licensing makes the weights available for commercial reuse, including by U.S. integrators and downstream developers.
Read at Pandaily ↗
Shanghai AI Lab and university partners developed a high purity KrF photoresist resin using an AI driven synthesis platform that achieves batch consistency metrics meeting semiconductor manufacturing standards, per Pandaily. The team has entered customer validation with domestic photoresist maker Hengkun New Materials. KrF photoresist is one of the chipmaking consumables Chinese fabs have long imported from Japanese and Western suppliers. Beijing has prioritized domestic alternatives under its semiconductor self reliance push.
Read at Pandaily ↗
Moore Threads, a U.S. sanctioned Chinese GPU maker, and simulation platform Guangyun Intelligence announced a strategic partnership to build a high confidence synthetic data solution for embodied AI, combining Moore Threads' domestic GPU compute with Guangyun's in-house simulation platform, per Pandaily. The partnership targets sovereign compute and synthetic training data for Chinese robotics developers operating without access to Nvidia accelerators. Synthetic data lets developers scale training data without physical fleet collection.
Read at Pandaily ↗
IV.Industry & Market Watch
Anduril Industries, the defense technology firm the New York Times calls a maker of AI backed weapons, said Wednesday it raised $5 billion at a $61 billion valuation, double a June 2025 round at $30.5 billion. Thrive Capital and Andreessen Horowitz led the latest round, Reuters reported. The funding round comes as defense tech investment expands amid the ongoing U.S.-Iran war. Anduril more than doubled revenue to $2.2 billion in 2025 and nearly doubled its workforce.
Read at NYT ↗ • Read at Reuters ↗
Anthropic is in early talks to raise at least $30 billion at a valuation of approximately $950 billion, up from $380 billion in its previous round, the New York Times reported. The Financial Times had earlier reported that the proposed round would fund a major expansion in Anthropic's computing capacity that could lift its valuation toward $1 trillion, per a Reuters summary of the FT story.
Read at NYT ↗ • Read at Reuters ↗
CME Group is launching a futures market for AI computing power, with prices based on Silicon Data's GPU rental benchmarks, SiliconANGLE reported. Silicon Data CEO Carmen Li said GPU markets have historically lacked standardized reference pricing, framing the new contracts as a way for AI builders, cloud providers and investors to hedge GPU prices. Buyers will be able to lock in prices for cloud-based GPU capacity, hedging against price moves as data center operators absorb global GPU supply.
Read at SiliconANGLE ↗
NV Energy will stop supplying power to Liberty Utilities, the company serving roughly 50,000 California-side Lake Tahoe customers, after May 2027 as the Nevada utility redirects capacity to Northern Nevada data centers, Fortune reported. NV Energy supplies about 75% of Liberty's power, with the remainder from Liberty-owned Nevada solar facilities. Liberty has asked the California Public Utilities Commission to authorize an expedited request for proposals for replacement energy beginning June 1, 2027. Residential rates have surged about 77% since late 2022, approaching twice the national average, Bloomberg reported separately.
Read at Bloomberg ↗ • Read at Fortune ↗
V.Global & Geopolitics
Japan's three largest banking groups (Mitsubishi UFJ, Sumitomo Mitsui and Mizuho) are set to gain access to Anthropic's Mythos model as soon as the end of May, Bloomberg reported, citing Nikkei. U.S. Treasury Secretary Scott Bessent informed the banks of the move at a Tokyo meeting Tuesday. Japanese Finance Minister Satsuki Katayama said after meeting Bessent that Tokyo will establish a public-private working group this week on Mythos related cybersecurity risks to the Japanese financial system, with the first meeting expected Thursday, Reuters reported. The move would mark the first time a Japanese company has been granted Mythos access.
Read at Bloomberg ↗ • Read at Reuters ↗
South Korea's Ministry of Science and ICT plans to announce countermeasures against AI powered cyberattacks as early as late May, officials said Tuesday, after what they called the "Mythos shock" from Anthropic's preview of advanced vulnerability finding capabilities, UPI reported. Science and ICT Minister Bae Kyung-hoon said the country must build a security system capable of defending against AI with AI. The ministry and the Korea Internet & Security Agency recently used public AI models to simulate an attack on corporate services and found seven vulnerabilities in about 10 minutes. South Korea has expressed interest in joining Anthropic's Project Glasswing partnership for defensive cybersecurity work.
Read at UPI ↗
South Korean President Lee Jae Myung clarified that policy chief Kim Yong-beom's reference to a citizen dividend concerned distribution of excess tax revenue generated from AI sector profits, not a redistribution of company profits, in a post on X, Bloomberg reported. Lee said some media outlets had edited the remarks and circulated "defamatory fake news" claiming Kim proposed reviewing a plan to distribute companies' excess profits to citizens. The clarification follows the original viral X post from Kim that prompted comparisons to a corporate windfall tax.
Read at Bloomberg ↗