AI Daily Digest
Your daily briefing on AI developments that matter to your business.
🎧 Listen to this digestSaturday, July 18, 2026
Government & Policy
3White House Moving to Control Which Companies Can Access Frontier AI Models
The Trump administration is asserting new authority over the rollout of frontier AI models, effectively creating a clearinghouse that would greenlight which companies and entities can access the most powerful new AI systems. Sources told CNBC the arrangement would put the White House in charge of approving access, though an official described any government engagement as 'voluntary.' The move represents a significant shift in power away from AI labs, which have historically controlled their own release timelines.
CNBCEU Tech Sovereignty Package Targets Supply Chain Resilience and Cyber Hardening Across Critical Sectors
The European Commission's June 2026 'tech sovereignty' package includes legislation requiring designated critical entities to implement broad resilience measures covering not just cyberattacks but also physical threats, natural hazards, and insider risks. The package aims to reduce European dependency on external tech suppliers while building domestic AI and cloud capacity. Analysts say it represents the EU's most comprehensive push yet to insulate its digital infrastructure from geopolitical disruption.
Global Policy WatchGoogle DeepMind CEO's Industry-Led AI Safety Proposal Draws Criticism Over Self-Regulation Concerns
Demis Hassabis has proposed an industry-funded AI safety standards body modeled on FINRA, the financial industry's self-regulatory organization, to oversee frontier AI risks in the U.S. Critics argue the proposal lacks independent oversight, clear legal authority, and meaningful safeguards against conflicts of interest. The debate echoes broader tensions over whether the AI industry can police itself or requires independent government regulation.
The Economic TimesMajor AI Players
2Anthropic in Early Talks to Lease Up to $10B in AI Compute From Meta
Anthropic is in preliminary negotiations to lease computing power from Meta in a deal that could reach $10 billion over two years, according to multiple reports. The arrangement would effectively position Meta as a cloud computing provider competing with Amazon, Microsoft, and Google — a move Meta CEO Mark Zuckerberg had publicly floated in May. The deal follows Anthropic's earlier compute agreement with SpaceX's Colossus data center and reflects the intense competition for Nvidia GPU capacity.
CNBCNational Data Center Protests Planned at 125 U.S. Locations as Public Backlash Against AI Infrastructure Grows
Opponents of rapid AI data center construction are coordinating protests at more than 125 locations across the U.S. on Saturday, marking the first nationally organized effort to push back against the AI infrastructure boom. A June Reuters/Ipsos poll found only a third of Americans approve of the current pace of data center construction, and just 14% would support one being built in their community. The backlash crosses ideological lines and is increasingly influencing local politics in states like New York, which already imposed a construction moratorium.
ReutersGlobal Developments
3Kimi K3 Stock Market Shockwave: Nasdaq Falls 1.5% as Chinese Open-Weight Model Rivals U.S. Frontier AI
Moonshot AI's Kimi K3 — a 2.8 trillion parameter open-weight model — rattled global markets Friday, sending the Nasdaq down 1.5% and Taiwan's stock index down more than 6% on fears that Chinese AI competition could undermine the U.S. AI spending thesis. Third-party evaluators ranked K3 third globally on overall capability, behind only Anthropic's Claude Fable 5 and OpenAI's GPT-5.6, and first in frontend coding — the first time a Chinese model has topped that category. The model is priced at roughly half the cost of GPT-5.6 and will be released as fully open-weight on July 27.
AP NewsWeRide Recognized as Only Autonomous Driving Company in China's Top 10 AI International Cooperation Cases
China's National Development and Reform Commission named WeRide's autonomous vehicle deployment in the UAE as one of the top 10 examples of Chinese AI technology benefiting the world, announced at WAIC 2026. WeRide was the sole autonomous driving company on the list, with its UAE operations cited as a benchmark for internationalizing Chinese AI innovation. The recognition underscores China's strategy of using real-world overseas deployments to build credibility for its AI exports.
Business Insider MarketsXi Jinping Makes First In-Person Appearance at WAIC, Frames WAICO Alliance as Counter to U.S. AI Dominance
Chinese President Xi Jinping appeared in person at the World Artificial Intelligence Conference in Shanghai, calling AI development a 'symphony of international cooperation' rather than 'a solo performance by a single country' — a clear jab at U.S. leadership. His remarks accompanied the formal launch of WAICO, a 29-nation AI cooperation alliance China announced Thursday, with UN Secretary-General Antonio Guterres attending the launch. Analysts say Beijing intends to use WAICO to shape global AI governance standards in its favor.
GizmodoRobotics
5mimic Robotics Unveils Vertically Integrated Physical AI Platform Including Robotic Hand, Exoskeleton, and Training Software
Swiss startup mimic Robotics has introduced the mimic hand M1, the mimic wearable U1 exoskeleton, and a software infrastructure layer that together form what it calls a 'vertically integrated physical AI platform.' The system is designed so the human hand acts as a fixed morphological reference across hardware, data collection, and AI model training — aiming to build a general-purpose robotics foundation model. The approach bets on consistency between human and robot hand shape as the key to scalable dexterous manipulation.
Robotics & Automation News1X Unveils 25-Degree-of-Freedom Tendon-Driven Hands for NEO Humanoid, Pushing Dexterity Toward Human Level
Norwegian humanoid robotics company 1X has revealed new tendon-driven hands for its NEO platform featuring 25 degrees of freedom, designed to match near-human dexterity, grip strength, and tactile sensitivity. The hands are built to enable more sophisticated AI-driven manipulation tasks that previous robotic hand designs couldn't support. The announcement positions 1X as a serious competitor in the race to produce humanoids capable of handling unstructured real-world environments.
Robotics & Automation NewsAgility Robotics Opens 60,000 Sq Ft Fremont Facility Steps From Tesla's Optimus Factory
Agility Robotics has opened a new software and AI development hub in Fremont, California, just miles from the Tesla factory where Optimus robots are expected to begin manufacturing this year. The facility will serve as the primary site for training, testing, and advancing the AI capabilities of Agility's Digit humanoid, which is already deployed at Amazon, GXO, and Toyota warehouses. CEO Peggy Johnson is simultaneously guiding the company through a reverse merger expected to make it the first pure-play humanoid robotics public company later this year.
TechCrunchWeave Robotics Launches Isaac 1, a Wheeled Home Humanoid Designed for Everyday Household Chores
Weave Robotics has introduced Isaac 1, a wheeled home humanoid robot designed to fold laundry, make beds, tidy rooms, and put away toys, with deliveries planned for this fall. Users control and schedule tasks through a companion app, and the robot features camera shutters that make its active/inactive status visually clear — a privacy design choice unusual among consumer humanoids. The product builds on Isaac 0, which found early commercial use folding clothes in laundry services.
The Robot ReportKorean Startup Holiday Robotics Closes $103M Series A — Largest Single Robotics Round in Korea This Year
Seoul-based Holiday Robotics has closed a $103.4 million Series A round, the largest single funding round in Korean robotics in 2026 and a signal that South Korea's humanoid manufacturing ambitions are entering industrial scale. The broader Korean startup ecosystem pulled in over $120 million this week, concentrated in AI infrastructure, enterprise automation, and robotics. International capital, including from a Saudi-backed deep-tech fund, is increasingly flowing into the Korean AI and robotics sector.
WowtaleTransportation
3San Francisco Emergency Chief Called Waymo a 'Havoc' Maker on July 4 as Internal Records Reveal Mayor's Push for Tougher AV Standards
Newly released public records show San Francisco's emergency management director texted the mayor's top aide on July 4 that Waymo 'continues to wreak havoc on the streets' after its robotaxis stalled and blocked traffic following the Golden Gate Bridge fireworks. Mayor Daniel Lurie followed up with a formal letter to California transportation regulators calling for mandatory statewide standards governing how autonomous vehicles behave during major disruptions. The internal records add new detail to the political pressure building against Waymo and the broader AV industry in its home city.
San Francisco ChronicleZoox Recalls 105 Autonomous Vehicles Over Failure to Detect Heavy Smoke, Risking Emergency Response Interference
Amazon's self-driving car unit Zoox announced a recall of 105 autonomous vehicles after determining they may fail to detect heavy smoke, potentially blocking emergency personnel response. The recall comes one week after the top U.S. auto safety official warned self-driving car companies to address a 'clear pattern' of driverless vehicles interfering with first responders. The incident adds to a growing list of AV safety incidents drawing federal regulatory scrutiny.
Yahoo Finance / ReutersTesla Driver Exploits Dash Cam Loophole to Use Full Self-Driving Hands-Free, Raising Safety Concerns
A Tesla driver has publicly described using a low-tech method to fool the vehicle's driver monitoring system — displaying a static image of himself looking at the road on a screen in front of the inside dash cam — allowing him to use FSD Supervised without keeping his hands on the wheel. The workaround highlights ongoing vulnerabilities in driver attention monitoring systems that rely on camera-based detection rather than more robust physical sensors. The video has circulated widely, reigniting debate about whether current FSD monitoring is adequate.
Motor1Healthcare & Biotech
4KLAS 2026 Report: Clinicians Using Epic AI Score Higher on Efficiency But Less Than 25% Received Adequate Training
The KLAS Arch Collaborative 2026 report finds that clinicians using at least one AI tool within Epic score 7 percentage points higher on believing their EHR enables operational efficiency compared to non-users, with the biggest gains in order creation and shift summarization. However, fewer than 25% of AI-adopting clinicians report receiving adequate training on managing AI-generated content within their workflows. The report urges health systems to shift focus from deploying AI tools to enabling role-specific workflow integration.
HIT ConsultantAqemia and Sanofi Expand AI Drug Design Collaboration With New Therapeutic Target Nomination
French AI drug discovery startup Aqemia has expanded its multi-year partnership with Sanofi, with the pharma giant nominating a new therapeutic target and triggering an additional milestone payment. Aqemia uses its Qemi platform — a generative AI system informed by quantum-inspired physics — to design novel small molecules for targets with limited chemical data. The expansion signals growing confidence in AI-driven molecule design for difficult, first-in-class drug programs.
BioSpaceAbbott Discloses Cyberattack on Cancer Diagnostics Business Acquired From Exact Sciences for $21B
Abbott Laboratories confirmed unauthorized access to internal systems in its cancer diagnostics business, which includes Exact Sciences, acquired earlier this year for $21 billion. Abbott said the legacy Exact Sciences systems are separate from its own infrastructure and that no material business impact is expected, but the company declined to detail when the attack was discovered or what data was accessed. The incident is part of a wave of cyberattacks hitting major medtech companies including Stryker, Medtronic, Intuitive, and iRhythm.
MedTech DiveHemispheric Raises $52M to Scale Descartes, a 6-Billion-Parameter AI Brain Foundation Model Trained on 250,000 Hours of Neural Data
Hemispheric has closed a $52 million funding round to expand its Descartes AI platform, a foundation model trained over six years on more than 250,000 hours of brain activity data. The company aims to address the lack of objective diagnostic tests for the brain — the only major organ without them — and is pursuing FDA regulatory clearance while expanding government and pharma partnerships. Hemispheric has already presented its platform to FDA's Center for Devices and Radiological Health leadership.
MobiHealthNewsInvestment Noteworthy
5Basis AI Agent Platform for Accountants Reaches $1.15B Valuation as Accounting Firms Scale Adoption
Basis, an AI agent platform built specifically for accountants, has reached a $1.15 billion valuation and is seeing adoption at top accounting firms across the country. CEO Matt Harpe said the platform is designed to handle the kinds of repetitive, high-stakes workflows that define modern accounting practice, and the company has attracted backing from top venture capital firms. The valuation milestone puts Basis among the first accounting-specific AI unicorns.
Accounting TodayDatabricks Hits $188 Billion Valuation in New ~$3B Funding Round, Cementing Status as AI Infrastructure Cornerstone
Databricks announced a new funding round valuing the company at $188 billion — a 40% jump from its February valuation of $134 billion — with approximately $3 billion raised and the round expected to close later this summer. The company plans to use capital to expand its multi-AI governance tool, workplace AI agent, and serverless Postgres database built for AI agents. Analysts noted the raise looks 'almost quaint' compared to Anthropic's recent fundraising pace, where the valuation more than doubled in a single quarter.
TechCrunchEuropean VC Hits €44B in H1 2026 as AI Takes 60%+ of Deal Value, With Mega-Rounds Dominating
European venture capital reached €44 billion in H1 2026 — 27% ahead of 2025's full-year pace — but funding is concentrating in fewer deals, with AI startups capturing more than 60% of total deal value and mega-rounds over €100 million accounting for more than half of Q2 capital. London-based AI infrastructure startup Nscale raised €2.9 billion, the region's largest deal. The trend mirrors U.S. patterns where AI mega-rounds are inflating overall funding figures while smaller deal counts signal a tightening market for early-stage non-AI startups.
PitchBookNuclear Startup Valar Atomics in Talks to Raise $1B at $6B Valuation, Backed by Sequoia, After Powering Nvidia AI Chip
Valar Atomics, a three-year-old small modular reactor startup, is in advanced talks to raise approximately $1 billion at a $6 billion valuation with Sequoia expected to lead the round. The El Segundo-based company recently demonstrated its nuclear reactor providing power to an Nvidia AI chip — a milestone designed to show the technology's relevance to AI data center energy demands. The funding would make Valar one of the most highly valued nuclear energy startups in the world.
TechCrunchGeneral Compute Secures $400M Credit Facility Using Inference Chips as Collateral in Potentially Historic AI Finance Deal
AI inference startup General Compute has secured a $400 million credit facility from investment firm Upper90, with the deal potentially representing the first major financing transaction in history to use specialized AI inference chips as collateral. The company uses SambaNova chips to run open-source models faster and more cheaply than hyperscaler alternatives, positioning itself as a cost-effective infrastructure option for businesses wanting to avoid OpenAI or Anthropic pricing. The deal reflects growing investor recognition that inference infrastructure — not just model training — is a critical and bankable AI asset.
Zamin.uz / TechCrunch