★ Top Story
The VergeMicrosoft's Build 2026 keynote marked a decisive shift toward AI independence, unveiling its first in-house reasoning model (MAI-Thinking-1), a new OpenClaw-inspired personal assistant called Scout, and Project Solara — an Android-based OS redesigned around agents rather than apps. The announcements collectively signal that Microsoft, having renegotiated its OpenAI partnership, is now positioning itself as a full-stack AI competitor in its own right. Scout integrates deeply into Microsoft 365, while MAI-Thinking-1 represents the company's most ambitious model development effort to date.
Read the original →
The VergePresident Trump signed a new executive order creating a voluntary framework for AI companies to share frontier models with the federal government before public release, framing it as a measure to strengthen cybersecurity and promote secure innovation. The order is notably narrower than earlier drafts, after the AI industry pushed back against mandatory oversight requirements. Critics, however, warn that the administration's capacity to meaningfully evaluate these models has been severely weakened by DOGE-driven cuts to federal security teams.
Read more →
Ars TechnicaThe UK Competition and Markets Authority has imposed new conduct rules on Google, requiring it to display clearer attribution links within AI Overviews and to provide website publishers with a genuine opt-out mechanism for AI Search features. The ruling is significant because Google had previously argued that users don't want to see many sources in AI-generated answers — a position regulators rejected. The opt-out tool will be piloted in the UK before a planned global rollout, setting a potential precedent for other markets.
Read more →
TechCrunchAlphabet completed an $85 billion stock offering — the largest in corporate history — to fund the continued buildout of Google's AI infrastructure and services. The record-breaking raise reflects extraordinary investor appetite for AI-linked equities at a moment when the sector is absorbing massive capital expenditures. Analysts say the signal is as much about market sentiment as it is about Alphabet's specific roadmap.
Read more →
TechCrunchAnthropic is scaling its most capable model, Claude Mythos, to 150 organizations across more than 15 countries as part of Project Glasswing, its security vulnerability program targeting critical infrastructure sectors including power, water, healthcare, and communications. The initiative is designed to harden systems that, if compromised, could affect up to 100 million people. The expansion represents one of the most ambitious deployments of a frontier AI model in high-stakes infrastructure environments to date.
Read more →
TechCrunchOpenAI released six specialized plug-ins for its Codex platform, each tailored to a distinct professional role: data analytics, creative production, sales, product design, equity investing, and investment banking. Each plug-in bundles relevant integrations, instructions, and contextual knowledge to let Codex approximate the output of a domain-specific knowledge worker. The move represents OpenAI's most direct push yet into automating white-collar workflows across industries.
Read more →
The VergeHands-on testing of Google's new Gemini-powered personal agent, Spark, revealed that the system proactively surfaces intimate personal details — including a user's dog's name and a spouse's first name — drawn from data across their Google account, without being explicitly asked. Reviewers found the capability simultaneously impressive and unsettling, raising pointed questions about the trade-off between AI usefulness and personal privacy. The experience has reignited debate about whether the promise of a truly helpful AI assistant is inseparable from deep, potentially uncomfortable surveillance of users' lives.
Read more →
TechCrunchMeta has made its AI agent for WhatsApp Business available worldwide, enabling companies to automate customer interactions at scale through the messaging platform. Businesses will be charged based on token usage, a pricing model that aligns costs with actual AI consumption. The global launch comes as Meta races to close the gap with rivals in enterprise AI, with WhatsApp's massive international user base offering a significant distribution advantage.
Read more →
Ars TechnicaA detailed look inside Meta reveals a company under pressure to close a widening gap with AI leaders like Google, OpenAI, and Anthropic, despite massive infrastructure investments and the open-source release of its Llama model family. Internal doubts persist about whether Meta's organizational culture and research priorities are well-suited to competing at the frontier of AI development. The report paints a picture of a company that has the resources but is struggling to translate them into decisive competitive advantage.
Read more →
Ars TechnicaGoogle's June Android feature update introduces on-device AI detection for spoofed and deepfake phone calls, targeting a growing wave of scams in which fraudsters impersonate family members, employers, or authority figures using AI-generated voices. The feature works by analyzing call audio in real time and alerting users when a call shows signs of AI manipulation or number spoofing. The rollout reflects how AI is increasingly being deployed both as a scam tool and as a countermeasure against those same threats.
Read more →
TechCrunchAI app-building platform Lovable has signed a multiyear agreement with Google Cloud that will quintuple its cloud footprint and expand its access to Anthropic's Claude models. The deal signals strong growth momentum for Lovable and reflects the increasingly competitive market for AI infrastructure partnerships, where cloud providers are aggressively locking in fast-growing AI-native companies. The expanded Claude access also highlights the deepening relationship between Google Cloud and Anthropic.
Read more →
TechCrunchObservability platform Coralogix has raised $200 million on the thesis that as AI agents move from experimentation into production, enterprises will urgently need tools to monitor their behavior, diagnose failures, and maintain operational reliability. The raise reflects a broader infrastructure investment wave targeting the "AI ops" layer — the unglamorous but critical plumbing that keeps agentic systems running safely at scale. Coralogix is competing in a rapidly crowding field as the agent deployment boom creates new categories of operational risk.
Read more →
The VergeAmazon is updating its in-app search bar to show AI-generated images of products as shoppers type their queries, initially covering clothing and home goods. The feature is designed to help users visually articulate what they're looking for and then surface real products that match the generated image. While Amazon frames it as a discovery aid, critics note the feature risks creating confusion between AI-invented visuals and actual purchasable inventory.
Read more →
NVIDIA BlogAt CVPR 2026, NVIDIA announced a suite of new physical AI agent skills and research tools aimed at accelerating development across autonomous vehicles, robotics, and vision AI systems. The releases address the full development workflow — from real-world scene reconstruction and edge-case scenario generation to policy training and evaluation — rather than just model performance in isolation. NVIDIA also showcased advances in generalized robot grasping that allow grippers to handle novel objects and tools without prior training on them.
Read more →
TechCrunchUber has imposed new limits on employee spending on AI tools after the company exhausted its full-year AI budget within the first four months of 2026, according to reports. The overspend came after Uber had actively encouraged staff to adopt and experiment with AI tools as broadly as possible. The episode highlights a growing tension at large enterprises between promoting AI adoption and managing the unpredictable costs that come with it.
Read more →
Ars TechnicaThe International Mathematical Union has endorsed a formal warning that AI systems capable of automated theorem proving and mathematical reasoning pose a serious long-term threat to the mathematics profession and to academic independence. The statement also raises concerns about the tech industry's growing financial influence over mathematical research, which critics say risks distorting research priorities toward commercially useful applications. The warning adds mathematics to a lengthening list of knowledge professions grappling with AI-driven disruption.
Read more →
Ars TechnicaA class action lawsuit filed in Seattle alleges that Amazon's Ring cameras use AI to scan and identify the faces of guests and passersby through its Familiar Faces feature, without obtaining meaningful consent from those individuals. The plaintiff argues that Ring users and the people they capture on camera are entitled to compensation under biometric privacy laws. The case could have broad implications for the legality of AI-powered facial recognition deployed in consumer home security products.
Read more →
The VergeFacing mounting public and regulatory scrutiny over the environmental footprint of its AI data center expansion, Google has published five specific commitments around water use, including a pledge to replenish more water to local communities than its facilities consume. The announcement is part of a broader effort by the company to get ahead of backlash as AI infrastructure buildouts strain water resources in communities across the US. Environmental advocates have welcomed the commitments but are calling for independent verification of the targets.
Read more →No stories from this source today.