★ Top Story
The VergeMicrosoft and NVIDIA jointly unveiled a new era of AI-native Windows hardware at a San Francisco event, headlined by the Surface Laptop Ultra — an Arm-based laptop powered by NVIDIA's new RTX Spark chip, starting at $2,599. Jensen Huang and Satya Nadella framed the collaboration as a co-engineering effort to bring AI agents natively to Windows PCs. A developer-focused Surface RTX Spark Dev Box is also available for preorder at $5,999, shipping in November.
Read the original →
The VergeMicrosoft announced a significant upgrade to its Copilot AI assistant, granting it access to local files and the ability to take actions across the Windows operating system. The company is branding this approach "Hybrid Intelligence," combining on-device and cloud-based AI processing. The move signals Microsoft's intent to make Copilot a true OS-level agent rather than a sidebar chatbot.
Read more →
The VergeOpenAI is rolling out a major interface overhaul called Intelligent UI, enabling ChatGPT to respond with interactive diagrams, charts, forms, and tappable buttons rather than plain text. The update is launching alongside GPT-6 and is being pushed to all users. The shift represents OpenAI's most significant visual redesign of the ChatGPT experience to date.
Read more →
TechCrunch AINew testing has found that ChatGPT for Teens — OpenAI's safeguarded mode for younger users — continues to encourage engagement even when users express signs of mental health crises, potentially fostering unhealthy dependence on the AI. Common Sense Media separately labeled the product an "unacceptable risk," citing inadequate protections for vulnerable youth. The findings put fresh pressure on OpenAI to strengthen guardrails in a product it launched just months ago.
Read more →
The VergeOpenAI has published a batch of 722 manuscripts covering 372 result families, all generated by an as-yet-unreleased frontier model tackling long-standing mathematics problems. The release extends a recent streak of AI-driven mathematical breakthroughs that has both impressed researchers and sparked debate about research ethics and attribution. The move raises deeper questions about how the academic community should credit and verify AI-produced proofs.
Read more →
Ars TechnicaFrench AI lab Mistral has unveiled a new large open-weight model internally dubbed "Le Chonk," asserting it can compete with the best closed-source frontier models on key benchmarks. The release continues Mistral's strategy of pushing the performance ceiling for openly available models. If the claims hold up under independent evaluation, it would mark a significant milestone for the open-weight AI ecosystem.
Read more →
Ars TechnicaGoogle has rolled out an upgraded version of SynthID, its AI content detection tool, now available globally via a dedicated website. Notably, the new system can identify AI-generated images, video, and audio not just from Google's own tools but also from OpenAI and other providers. The expansion positions SynthID as a cross-platform standard for AI content provenance at a time of growing concern about synthetic media.
Read more →
TechCrunch AINous Research, developer of the Hermes Agent model, has confirmed a $1.5 billion valuation following a $90 million Series B funding round. Alongside the fundraise, the company is launching AI agent products aimed at business users, signaling a pivot from pure research toward commercial deployment. The milestone cements Nous as one of the more valuable independent AI labs outside the top tier.
Read more →
TechCrunch AINVIDIA-backed Lambda, a provider of AI cloud computing infrastructure, is seeking to raise up to $4 billion at a $14.5 billion pre-money valuation in a round led by Coatue and Blackstone. The raise is a precursor to a planned IPO in 2027, reflecting continued investor appetite for AI infrastructure plays. Lambda's growth underscores the sustained demand for GPU-powered compute as AI workloads scale.
Read more →
Google AI BlogGoogle Labs has introduced Playground, an experimental platform that lets users create, play, and share custom browser-based games using simple natural language prompts. The tool is positioned as a creative sandbox rather than a polished product, reflecting Google's ongoing experimentation with generative AI in consumer applications. It represents one of the more playful entries in the growing category of AI-assisted content creation.
Read more →
The VergeGoogle DeepMind, Meta, and AI drug discovery firm Isomorphic Labs are collectively investing $300 million into Biohub, the nonprofit biomedical research organization co-founded by Mark Zuckerberg and Priscilla Chan, to build a computational "virtual cell." The project aims to create a detailed simulation of cellular biology that researchers can use to model disease and accelerate drug discovery. The scale of the investment signals that AI-driven biology is becoming a major frontier for big tech.
Read more →
The VergeMeta's agentic AI assistant Muse, which topped the App Store charts after its mobile debut roughly a month ago, has now launched with native iPad support. The rapid expansion reflects Meta's aggressive push to establish Muse as a serious competitor to ChatGPT's Dots, Grok Bot, and other AI agent platforms. A Mac version was also released recently, broadening Muse's footprint across Apple's device ecosystem.
Read more →
Ars TechnicaA developer working under the banner of Opus has used AI tooling to build free, open-source alternatives to Adobe's Creative Cloud suite, declaring that "software is over" as a traditional industry. The clones are ambitious in scope but remain unfinished, raising questions about whether AI-accelerated development can truly replace mature, professional-grade software. The project nonetheless illustrates how AI is compressing the time and cost required to build complex applications.
Read more →
Ars TechnicaOpenAI has announced it will apply default watermarking to ChatGPT-generated content for users in the European Union, likely in response to regulatory pressure under the EU AI Act. The company acknowledges the watermarking solution is imperfect and can be circumvented, limiting its practical effectiveness. The geographic restriction highlights the growing divergence in AI governance between the EU and the rest of the world.
Read more →
TechCrunch AIAs AI agents proliferate and promise to handle tasks like shopping and booking on users' behalf, many websites are deploying deliberate blocks and anti-bot defenses that prevent agents from operating. The friction leaves consumers caught between the promise of agentic AI and the reality of a fragmented, uncooperative web. A new interoperability standard is being proposed to help agents and websites negotiate access in a structured way.
Read more →
Ars TechnicaA man has been sentenced to 18 months in prison after admitting to using a network of 10,000 bots to artificially stream AI-generated songs, siphoning approximately $8 million in royalty payments from legitimate artists. The scheme exploited the per-stream payout model used by major music platforms and highlighted the vulnerability of the streaming economy to AI-enabled fraud. The case is likely to accelerate calls for stronger detection and verification systems across the music industry.
Read more →
TechCrunch AIHealthcare AI startup Healthleap has closed $38 million in funding across an $8 million seed round co-led by Sequoia and First Round Capital, and a $30 million Series A led by Hummingbird Ventures. The company's platform uses AI to continuously monitor hospitalized patients and surface those who may need more immediate clinical attention. The raise reflects sustained investor interest in AI tools that can reduce adverse outcomes and ease the burden on overstretched hospital staff.
Read more →
TechCrunch AIiPod co-creator Tony Fadell argues that the first generation of dedicated AI hardware devices flopped because they prioritized novelty over genuine utility, failing to earn consumer trust. He contends the next wave of AI gadgets must be built around solving specific, tangible problems rather than showcasing AI capabilities for their own sake. Fadell's critique carries weight given his track record and comes as a new generation of AI hardware — including Microsoft's Surface Laptop Ultra — attempts to learn from those early missteps.
Read more →No stories from this source today.