★ Top Story
The VergeThe Trump administration abruptly ordered Anthropic to cut access to its newest models — Fable 5 and Mythos 5 — for all foreign nationals, including users inside the US and Anthropic's own employees, citing export control rules. The move forced a broad service blackout that caught the company off guard and exposed how vaguely written AI export regulations can trigger sweeping, unintended consequences. Anthropic has been fighting to restore access while the industry watches closely for what the enforcement precedent means going forward.
Read the original →
TechCrunchAt the G7 summit, French President Macron and Indian PM Modi voiced alarm that the US could unilaterally cut off access to American AI systems overnight — a fear the Anthropic export-control blackout has now made concrete. The episode is accelerating conversations among allied governments about AI sovereignty and the risks of deep dependence on US-based frontier models. It may also spur investment in domestic or multilateral AI alternatives as a hedge against future disruptions.
Read more →
TechCrunchDespite the chaos of the export-control shutdown, spending data from Ramp suggests Anthropic's enterprise adoption has been growing strongly, and the public clash with the administration may be reinforcing its reputation as an independent, safety-focused lab. The dynamic is an unusual one: a government crackdown that was meant to constrain the company appears to be functioning as a brand differentiator. Analysts are watching whether the goodwill translates into durable enterprise contracts or fades once the news cycle moves on.
Read more →
TechCrunchA new Pew Research study finds that just 16 percent of Americans believe AI will be a net positive for society, even as chatbot usage has surged — with 49 percent now reporting occasional use, up from 33 percent in 2024. The gap between widespread adoption and deep public skepticism is striking: 63 percent of Americans think the technology is advancing too fast. The findings present a significant trust deficit that the industry has yet to meaningfully address.
Read more →
Ars TechnicaA new analysis argues that AI models with advanced offensive cybersecurity capabilities are on an inevitable trajectory toward mainstream availability, regardless of what any single government or company does to slow them down. The piece examines how the proliferation of capable open-weight models and the global race dynamics among labs make unilateral restrictions increasingly ineffective. The conclusion is sobering: policy frameworks built around preventing dangerous models from existing may need to shift toward managing their consequences instead.
Read more →
TechCrunchNEA investor Tiffany Luck says the "tokenmaxxing" era — where companies pushed AI usage to its limits — has collided with reality as budgets blow out faster than expected. Uber reportedly burned through its annual AI budget in just a few months, Meta shut down an internal AI usage leaderboard, and some firms have quietly trimmed Claude licenses. Luck argues the industry is entering a more disciplined phase where demonstrable ROI, not raw usage volume, will determine which enterprise AI deployments survive.
Read more →
The VergeGoogle's first new smart speaker in six years opens for preorder today and ships June 25th, replacing the rigid command-and-response model of Google Assistant with more fluid, conversational Gemini interactions. The $99 device is positioned less as an audio product and more as a Gemini access point for the home, signaling Google's bet that generative AI can revive a stagnant smart speaker category. Hardware specs are unchanged from the announcement nine months ago, suggesting the AI software layer is the real product.
Read more →
The VergeMidjourney CEO David Holz revealed The Midjourney Scanner, a ring-of-sensors ultrasound device capable of producing full-body scans — a dramatic pivot from the AI image generation the company is known for. Holz also announced plans to open a San Francisco spa as a consumer venue for the technology, framing it as a health and wellness offering. The move signals Midjourney's ambition to expand well beyond generative imagery into physical AI-powered hardware and medical diagnostics.
Read more →
Ars TechnicaNVIDIA has demonstrated a self-improvement loop for physical robots in which teams of AI coding agents autonomously generate and refine robot training programs, enabling machines to learn practical tasks like installing GPUs and cutting zip ties without direct human instruction. The system represents a meaningful step toward robots that can bootstrap their own skill acquisition, reducing the bottleneck of hand-crafted training data. It also underscores NVIDIA's deepening push into physical AI as a complement to its data center dominance.
Read more →
Google AI BlogResearch published in Nature shows that Google's AMIE conversational AI system performed on par with primary care physicians when managing complex chronic disease cases in a controlled study. The findings add to a growing body of evidence that AI can handle substantive clinical reasoning tasks, not just administrative or triage functions. Google is careful to frame AMIE as a tool to augment physicians rather than replace them, but the Nature publication gives the work significant scientific credibility.
Read more →
TechCrunchSnap debuted its long-awaited Specs AR glasses at $2,195, positioning them as a premium computing-in-the-world device that CEO Evan Spiegel says has been in development for over 12 years. Investors were unimpressed, sending the stock lower as questions mounted about the addressable market for a fashion-forward wearable at that price point. The launch puts Snap in direct competition with Meta's Ray-Ban glasses and Apple's Vision Pro, both of which have struggled to achieve mass-market traction.
Read more →
TechCrunchOdyssey has secured a funding round valuing it at $1.45 billion, with Amazon among the backers, as investor appetite for AI world models — systems that learn persistent, physics-grounded representations of environments — heats up beyond the LLM wave. World models are seen as a critical missing layer for robotics, autonomous vehicles, and simulation-heavy industries. The round cements Odyssey as one of the defining startups in what many researchers consider the next frontier of AI architecture.
Read more →
Ars TechnicaUK retail giant Tesco has filed court documents accusing Broadcom of hiking VMware licensing prices by approximately 175 percent and is now migrating 40,000 server workloads off the platform in response. The case is one of the most high-profile examples yet of enterprise backlash against Broadcom's post-acquisition VMware pricing strategy, which has drawn complaints from customers across industries. The migration effort will take years and significant resources, illustrating the real cost enterprises face when a critical infrastructure vendor dramatically changes its commercial terms.
Read more →
Ars TechnicaThe US Department of Defense has publicly claimed it is using generative AI to draft reports required by Congress, and says 1.5 million military personnel are now using AI tools. The disclosure raises immediate questions about oversight, accuracy, and whether AI-generated official reports meet the intent of congressional mandates. Critics argue that automating government accountability documents with AI introduces new risks of hallucination and reduced institutional accountability at exactly the wrong level of the chain of command.
Read more →
TechCrunchPramaana Labs has closed a $27 million seed round led by Khosla Ventures to apply formal verification techniques — mathematical proofs of correctness — to AI outputs in high-stakes domains like law, drug discovery, and tax preparation. The startup is betting that as AI moves deeper into regulated industries, probabilistic accuracy will no longer be sufficient and provably correct outputs will command a significant premium. The raise is one of the largest seed rounds in the AI reliability space and signals growing investor conviction that "trustworthy AI" is a distinct and fundable product category.
Read more →No stories from this source today.