AI Workers Call for a Pause, Nvidia Bets Half a Trillion, and Europe's New AI Law Is Live

Today's brief covers a rare act of industry self-restraint, a detailed forensic report on an AI-agent cyberattack, a new cybersecurity coalition with some conspicuous absences, Alphabet's mixed earnings quarter, Europe's updated AI regulation entering force, and Nvidia's potential role as financier of the largest infrastructure project in corporate history.

1,100 AI Insiders Sign Letter Demanding Governments Slow Frontier Development

More than a thousand AI workers — including executives and researchers from Anthropic, OpenAI, Google DeepMind, and Meta — have signed an open letter titled "Pacing the Frontier," calling on governments to deliberately slow the development of frontier AI systems. The petition represents a rare moment of public pressure from inside the industry itself: the very people building the most powerful AI in the world asking authorities to constrain their own work.

The letter urges an internationally coordinated approach to governing the pace of AI development, arguing that current deployment timelines outstrip humanity's ability to assess or manage the risks involved. It stops short of demanding a full moratorium but explicitly calls for mechanisms that would slow progress when safety assessments are not met. The fact that employees of competing frontier labs signed the same letter signals a level of cross-industry concern that has rarely been seen at this scale — and puts pressure on policymakers who have largely deferred to the industry on questions of timing.

Hugging Face Forensic Report: 17,600 Attacker Actions — and Why Chinese AI Defended the Network

Hugging Face has released a detailed forensic report on the breach of its systems by what appeared to be a compromised OpenAI agent between July 9 and 13. The report documents 17,600 distinct attacker actions during the four-day intrusion, offering one of the most granular public accounts of an AI-agent-based cyberattack on record.

What made the incident especially notable was Hugging Face's choice of defensive tool: the company turned to GLM-5.2, a model developed by Z.ai, a Chinese AI firm, to help analyze and counter the attack. The reason given was that US closed-source models — including those from OpenAI and Anthropic — were considered too safety-restricted to handle the aggressive, adversarial reasoning required for forensic defense work. That explanation has set off a pointed debate: if frontier Western AI models are too cautious to be useful in high-stakes defensive security contexts, what does that mean for national and corporate cybersecurity strategies that increasingly depend on AI?

Nvidia Launches Open Secure AI Alliance — OpenAI, Google, and Anthropic Notably Absent

Nvidia has announced the Open Secure AI Alliance, a coalition of more than 30 companies — including Microsoft, IBM, SpaceX, Hugging Face, and the Linux Foundation — focused on addressing the growing threat landscape around AI infrastructure. The alliance plans to develop open frameworks, shared threat intelligence, and collaborative defenses tailored to AI systems and the data centers that run them.

Conspicuously absent from the founding membership are OpenAI, Google DeepMind, and Anthropic — the three largest closed-model AI labs. Their exclusion from a security alliance led by one of the industry's dominant hardware suppliers is already drawing scrutiny. The timing is pointed: the announcement comes days after a highly publicized intrusion involving an OpenAI agent, and critics are asking why the companies whose models are most widely deployed — and most commonly targeted — are not part of a coalition designed to protect them.

Alphabet Q2: 24% Revenue Growth, Negative Cash Flow, and a 6% Share Drop

Alphabet reported second-quarter revenue of $119.8 billion, a 24% year-over-year increase, with growth driven partly by equity investment gains in addition to its core advertising and cloud businesses. Despite the strong top-line number, investors sent shares down roughly 6% after the company disclosed a significant increase in planned AI infrastructure spending.

Full-year capital expenditure guidance was raised to between $195 billion and $205 billion. Actual cash flow for the quarter turned negative at $5.9 billion — a sharp contrast with the headline revenue figures. The market reaction captures a tension that has become a recurring theme across the major AI players: investors remain enthusiastic about AI's revenue potential but are growing increasingly wary of the capital intensity required to compete at the frontier. For Alphabet, the math currently requires betting enormous sums on infrastructure whose returns are still playing out.

EU AI Omnibus Regulation Enters Force — Key Deadlines Run Through 2028

The EU AI Omnibus Regulation has entered into force, bringing a revised and streamlined version of the original AI Act into legal effect. The same week saw the publication of EN 18286, the first European AI standard, released July 22 as a technical benchmark that will underpin compliance assessments across the continent.

Implementation is staggered over the next two years. Rules governing high-risk applications — covering biometric identification systems, critical infrastructure, and employment-related AI — take effect on December 2, 2027. Broader obligations follow through 2028. The Omnibus version was designed to address widely-cited complexity concerns in the original AI Act, aiming to reduce compliance burden without weakening core safety and transparency requirements. For any global company with European customers or operations, the compliance clock is now officially running.

Nvidia Eyes $250B Financing Role in OpenAI's $500B Ohio Data Center

Nvidia is reportedly in talks to provide a $250 billion financing backstop for OpenAI's planned 10-gigawatt data center campus in Piketon, Ohio — a facility being developed on a former uranium enrichment site. The total campus cost is projected to reach as high as $500 billion, which would make it the most expensive single corporate infrastructure project in history.

In a related but separate development, Nvidia is also reportedly negotiating to finance up to $350 billion in chip purchases, potentially connected to the same campus. Together, these figures suggest Nvidia is positioning itself not just as a hardware supplier but as a financial infrastructure partner at a scale that has no precedent in the technology industry. The Piketon campus, if completed at the projected 10-gigawatt scale, would draw as much power as several mid-sized cities combined and would represent a defining bet on AI's long-term industrial footprint in the United States.