NeuralStackModal Labs Leads Leaner Week for U.S. Startup Funding With $750M Round
Modal Labs raised $750 million in a Series D, leading a week in which only two U.S. startup financings reached the multiple-hundreds-of-millions range. Icon followed with $200 million, while biotech, healthcare AI, insurance and semiconductor startups also secured sizable investments. The mix suggests investors remain willing to fund AI infrastructure and specialized technology, but are showing more restraint on outsized rounds.
Newsom Orders AI Kill-Switch Review as 2028 Ambitions Take Shape
California Governor Gavin Newsom has ordered independent verification of AI models and created an expert committee to examine emergency shutdown mechanisms for potentially harmful AI agents. The move revisits safeguards he rejected in 2024 while he considers a 2028 presidential campaign and prepares to leave office in January. California’s next steps could influence how the United States balances frontier-AI safety, industry innovation and federal reliance on self-regulation.
Nearly 3,000 Californians Urge Tougher Oversight of AI
Nearly 3,000 Californians offered recommendations on how the state should develop, use and regulate artificial intelligence. The Newsom administration’s Engaged California report highlights accountability for AI harms, the technology’s resource demands and safeguards for high-stakes uses. The feedback does not create new rules, but it could shape California’s next phase of AI policy as the state weighs stronger oversight.
Microsoft President Calls for Global AI Rules, Warning Milk Is Safer Regulated
Microsoft president Brad Smith says AI currently faces fewer safety rules than milk and washing machines. He is calling for emergency brakes, mandatory incident reports, victim notifications and independent evaluation of powerful systems. The proposal puts Microsoft at odds with President Donald Trump’s preference for industry self-regulation and raises pressure for binding international standards.
EU AI Omnibus Takes Effect, Delaying Key High-Risk AI Rules
The EU’s AI Omnibus entered into force on 27 July, simplifying parts of the AI rulebook while extending compliance timelines. Rules for high-risk systems under Annex III now apply from 2 December 2027, while high-risk AI embedded in physical products follows on 2 August 2028. The package broadens testing opportunities and eases obligations for smaller companies without removing safeguards for safety and fundamental rights.
Chrome’s 15-Year Web Legacy Points to an AI-Enhanced Future
Chrome is marking its 15th anniversary by outlining a developer-focused vision built around safer browsing, richer web capabilities, AI, and improved tooling. The Chrome team says Chromium’s role has expanded from supporting a browser to helping protect the web’s openness while working with browsers, standards groups, frameworks, and open-source contributors. Its next priorities could shape how AI experiences are delivered on the web and how efficiently developers build, debug, measure, and automate them.
Google, OpenAI and Nvidia Sign White House Super Intelligence Accord
Google, Anthropic, Meta, Nvidia, OpenAI and xAI executives signed a White House accord establishing four layers of AI controls and auditing. The agreement covers model monitoring during training and deployment, internal oversight teams, independent evaluators and board-level review. Its impact will depend on how consistently the companies apply these controls and develop shared safety standards.
Trump’s AI ‘hoax’ stance collides with growing bipartisan alarm in Congress
US efforts to regulate artificial intelligence are stalled as President Donald Trump rejects warnings about AI risks and dismisses calls for guardrails. Senior Democrats and several Republicans are urging Congress to delay its recess or convene a special session to debate safeguards, but divided proposals and Republican control in both chambers make swift action unlikely. The deadlock leaves AI companies largely to self-regulate even as lawmakers float ideas from federal kill switches to moratoriums on new data centres.
California Targets AI Kill Switch as Governance Pressure Builds
California Governor Gavin Newsom has ordered the state to develop emergency shutoff requirements for AI systems, marking one of the clearest state-level moves yet toward forced control mechanisms. The same week, the EU AI Office received its first formal GPAI systemic risk evaluations and lawmakers in Washington warned that broad federal preemption could leave consumers unprotected before any national replacement exists. The developments underscore how AI governance is moving from abstract policy debate into active enforcement, emergency planning, and jurisdictional conflict.
Senate AI Bill Collides With Fast-Rising Agent Economy and Patchwork Rules
U.S. lawmakers are racing to craft liability-focused rules for advanced AI just as consumer-facing agents flood the market with almost no federal guardrails. Cities and states from New York to California and Connecticut are stepping into the vacuum, experimenting with new frameworks for safety, employment, pricing and child protection. The emerging clash between frontier model oversight and the messy reality of AI agents will shape how risk, responsibility and innovation are balanced in the next wave of AI deployment.
EU AI Act and California Rules Flip From Foresight to Enforcement
A wave of long-planned AI transparency rules in Europe and California quietly turned into binding obligations in early August 2026. Regulators from Brussels to New Delhi and state officials in the US are moving ahead with sector-specific oversight even as broader US legislation stalls. The shift forces AI providers and deployers to treat policy calendars as operational constraints, not background noise.
Interpol Uses AI to Flag 126 Suspected Foreign Terrorist Fighters
Interpol says an AI-assisted sweep of jihadist imagery has identified 126 suspected foreign terrorist fighters so far. The five-day June operation used agents and scripts to sort more than 108,000 facial images down to a few thousand usable matches, with human officers verifying the output. It highlights how law-enforcement agencies are rapidly operationalising AI, even as governments and the public argue over how hard to rein it in.