AI Business · 02 Jul 2026

AI Business | Jul 2, 2026

Analysis

The June 12 Commerce/BIS order that pulled Anthropic's Fable 5 and Mythos 5 — the first US export controls applied to an AI model rather than the chips that train it — is now generating second-order costs on two fronts: Just Security argues the shift to ad hoc licensing is pushing India's defense-technology ecosystem toward architectural choices that complicate multi-administration agreements such as iCET and INDUS-X, while contributors in Lawfare note the ban drew sovereignty-branded open-weight releases within days — Z.ai's GLM-5.2, a Rio de Janeiro model built on Alibaba's Qwen, and Japan's Sakana "Fugu Ultra" — each marketed on freedom from US export risk.
As national controls fragment the market, the UN and its ITU are convening the AI for Good Global Commission on July 8 in Geneva, co-chaired by Salesforce's Marc Benioff and Rwandan President Paul Kagame and seating Amazon's Andy Jassy, Anthropic co-founder Jack Clark, Cohere's Aidan Gomez, Microsoft's Brad Smith, and Nvidia's Jensen Huang — an attempt to reconnect frontier-AI builders with global policymakers even as the same firms navigate divergent export and sovereignty regimes; the ITU puts the population the commission aims to reach at 2.2 billion without internet access.
The FTC's proposed policy — holding that AI firms training chatbots to avoid discriminatory responses may themselves violate Section 5 of the FTC Act, and that complying with Colorado's AI anti-discrimination law could breach federal law — opens a US domestic-regulation front distinct from export policy, arriving as the same Lawfare contributors argue Washington's leverage lies in regulating AI at home rather than restricting models abroad; public comment runs through July 31.
The compute substrate showed both a demand boom and a localization push on the same day: South Korea's June exports grew at their fastest pace since October 1978 on Samsung and SK Hynix AI-memory demand, while Chinese carmakers led by BYD — whose 7,000-engineer team unveiled the in-house Xuanji A3 autonomous-driving chip — accelerate domestic chip self-reliance, and Campbell, California startup Oxmiq raised $35 million from backers including MediaTek, Pegatron, and Samsung Catalyst Fund to license a single IP block combining GPU, CPU, and tensor engines, aimed at lowering the cost of building and running AI.
Commercialization of AI for science drew two data points as both frontier labs and pharma majors direct capability toward drug discovery: Japan's Takeda formed an AI drug-discovery partnership with InSilico worth up to $600 million, following Anthropic's launch this week of Claude Science, a dedicated workbench for pharmaceutical and biotech research.

POLICY & REGULATION

TRADELawfare | Will the New Export Controls Shake the Foundations of the U.S. AI Industry? | On June 12, the Commerce Department's Bureau of Industry and Security directed Anthropic to deny all foreign nationals access to its two most capable models, Fable 5 and Mythos 5, citing a possible "jailbreak" of the models' cyber guardrails — the first time the United States applied export controls to an AI model rather than the semiconductor chips that train it — and the company pulled the models for all users. Writing in Lawfare, Stanford's Alvin Graylin and Jon Rosenwasser of Four Corners Analysis noted the ban drew immediate sovereignty-branded responses: Chinese lab Z.ai released GLM-5.2 under a permissive MIT license the next day, Rio de Janeiro's city government published an open model built on Alibaba's Qwen, and Japan's Sakana released a "Fugu Ultra" model touted as offering "frontier capability without the risk of export controls." They argued that even after the Fable 5 ban was reversed, the episode left a lasting loss of trust, and that US defense contractor Palantir urged customers to pursue AI sovereignty through open-weight models.
TRADEJust Security | U.S. Export Control Unpredictability Is Testing the Limits of U.S.-India Tech Cooperation | Secretary of State Marco Rubio signed defense-cooperation agreements with India during a visit to New Delhi, including an Underwater Domain Awareness roadmap with Quad partners Japan and Australia, and said the two countries intend to move from a defense producer-buyer relationship to co-development of advanced military technology. Just Security argued that Washington's shift from a rules-based export-licensing regime to "ad hoc horse-trading" at a burdened Commerce Department is pushing India's defense-technology ecosystem toward architectural choices that will complicate implementing multi-administration commitments such as the 2023 iCET initiative and the INDUS-X and TRUST defense projects. The analysis noted that despite 50% tariffs that strained ties over the past year, Rubio called India one of the "most important strategic partners" in the world; the US designated India a "Major Defense Partner" in 2016 and granted license-free access to many dual-use technologies in 2018.
LEGAxios | Exclusive: UN launches "AI for Good" commission | The United Nations and its International Telecommunication Union are convening the AI for Good Global Commission, which will hold its first meeting on July 8 in Geneva. Salesforce CEO Marc Benioff and Rwandan President Paul Kagame will co-chair the body, whose members include ITU Secretary-General Doreen Bogdan-Martin, Estonian President Alar Karis, and policymakers from Kazakhstan, Namibia, Saudi Arabia, Singapore, and Nigeria. Tech leaders on the commission include Amazon CEO Andy Jassy, Anthropic co-founder Jack Clark, Cohere co-founder Aidan Gomez, Microsoft president Brad Smith, and Nvidia CEO Jensen Huang. Benioff told Axios the inaugural meeting will focus on strengthening AI infrastructure and accelerating AI's impact on health, education, food security, and disaster response; the ITU estimates 2.2 billion people still lack internet access.

PARTNERSHIPS & ENTERPRISE

PRODVentureBeat | Restaurants can now accept orders placed directly from ChatGPT and Claude thanks to Square's new integration | Square launched a ChatGPT app and a Claude plugin that let consumers discover restaurants and place orders directly inside the two AI platforms, allowing merchants to accept orders from users and their AI agents without technical setup. Square is not charging the marketplace commissions that third-party apps levy — DoorDash takes 15% to 30% and Uber Eats 20% to 30% on delivery — though it still applies its standard online-ordering fees of 3.3% plus $0.30 or 2.9% plus $0.30 for merchants on its Square Plus and Square Premium plans. The integration pulls from the live Square catalog to map items, pricing, modifiers, and stock in real time via the Order by Cash App plugin, with checkout completed in-chat or on the merchant's own ordering page.

GOVERNANCE & SAFETY

GOVReuters | US FTC says AI bias safeguards may run afoul of consumer law | The Federal Trade Commission said on Wednesday that AI companies whose chatbots produce responses reflecting "ideological objectives" may violate federal law, as part of a proposed policy on how it will apply its authority to the sector. The FTC said firms that train chatbots to avoid responses discriminating against specific groups could run afoul of Section 5 of the FTC Act, which bars unfair or deceptive practices, and that complying with a Colorado law aimed at preventing AI-driven discrimination in employment and other consequential decisions could itself violate the federal statute. President Donald Trump and other conservatives have accused AI chatbots of political bias; FTC Chairman Andrew Ferguson has previously used the agency's authority in cases addressing conservative grievances. The agency will accept public comment until July 31.

RESEARCH & MODELS

SCIUPDATEThe Wall Street Journal | Takeda, InSilico Strike AI Drug-Discovery Deal Worth Up to $600 Million | Japanese pharmaceutical major Takeda is forming an AI drug-discovery partnership with biotech firm InSilico with a potential value of up to $600 million, the Wall Street Journal reported, adding a major pharmaceutical partner to the growing set of AI drug-discovery arrangements.

Still in play: Anthropic's Claude Science launch (covered 1 Jul) continued to draw coverage, with MIT Technology Review recapping the workbench built to support pharmaceutical and biotech research.

COMPUTING & INFRASTRUCTURE

CHIPUPDATEThe Wall Street Journal | South Korean Exports Soar on AI Chip Boom | South Korea's exports rose in June at their fastest pace since October 1978, as the AI boom drove demand for chips made by Samsung Electronics and SK Hynix, according to official data cited by the Wall Street Journal. The report said the surge coincided with a strong run in South Korean equities.
CHIPReuters | Startup Oxmiq raises $35 million to build chip architecture to lower cost of AI | AI startup Oxmiq said it raised $35 million to build chip-design architecture and software intended to lower the cost of building and running AI applications, bringing its total raised to $60 million. CEO Raja Koduri, a former Intel chief architect and ex-AMD executive, said the company aims to collapse a graphics chip, a central processor, and a tensor engine into a single licensable block of intellectual property — "We would want to be the Arm of this next era," he said — and to enter the custom-chip market where Broadcom, Marvell, and MediaTek compete. The round, led by Samsung Catalyst Fund and Fudomo, included MediaTek and Pegatron Venture Capital; the Campbell, California company plans to hire more engineers.
CHIPFinancial Times | Chinese carmakers' hunger for chips boosts national self-reliance drive | Chinese carmakers are accelerating efforts to cut dependence on foreign chips, a step toward the chip self-sufficiency long sought by Beijing's industrial policy. BYD, the world's largest EV producer, in May unveiled the Xuanji A3, the first autonomous-driving chip designed by its 7,000-strong semiconductor team, with founder Wang Chuanfu saying the company can now supply all key chips for intelligent vehicles. Chinese automakers including Nio, Xpeng, SAIC, Changan, Great Wall Motor, Li Auto, and Geely are designing chips with AI functions and partnering with local developers such as Huawei, Horizon Robotics, and Black Sesame, though they still rely on TSMC, Samsung, and Germany's Infineon for high-end fabrication. UBS analyst Jimmy Yu said automotive applications could be a crucial growth driver for China's chip-design industry over the next three to five years; new EVs in China carry nearly twice the chips of combustion cars, worth up to $2,000 per vehicle.

Calendar

LEGJul 6–7 — UN Global Dialogue on AI Governance convenes, ahead of the ITU's AI for Good summit.
LEGJul 8 — UN/ITU AI for Good Global Commission holds its inaugural meeting in Geneva during the ITU's AI for Good Global Summit.
GOVJul 31 — Public-comment period closes on the FTC's proposed policy on AI chatbot bias and Section 5 of the FTC Act.

MARKETS

1 Jul 2026 close | Retrieved 1 Jul 22:00 UTC | Yahoo Finance

AI Infrastructure (1D) | CoreWeave 85.69 USD -13.9%

Indices (1D) | NASDAQ 26,040 -0.7% | SOX 13,353 -6.3%

Coverage: 1 Jul 01:00 – 2 Jul 01:00 UTC

§ Subscribe

Read AI Business every morning.

One click. Free during preview. Reply to refine or cancel.

Want a Beat tuned to your specific interests? Start here →