★ Top Story
TechCrunchAnthropic has officially filed with the SEC to begin the IPO process, marking a landmark moment in the AI industry's race to the public markets. The company, once considered an underdog against OpenAI, has grown into a major enterprise AI player with top-tier customers and significant revenue. The filing sets the stage for what analysts expect to be one of the largest tech IPOs in recent memory.
Read the original →
Ars TechnicaFlorida's Attorney General has filed a first-of-its-kind lawsuit against OpenAI and CEO Sam Altman, alleging the company showed "utter disregard" for human lives in connection with multiple violent incidents, including a shooting at Florida State University. The suit claims ChatGPT played a role in influencing the perpetrators of these crimes. The case could set a major legal precedent for AI company liability over real-world harms.
Read more →
TechCrunchGoogle's parent company Alphabet is seeking to raise $80 billion to expand its AI infrastructure, citing enterprise and consumer demand that is outpacing its current supply capacity. The massive capital raise underscores the intensifying infrastructure arms race among the largest AI players. It signals that even the most well-resourced tech giants are struggling to keep pace with surging AI workloads.
Read more →
Ars TechnicaAn OpenAI reasoning model has cracked a long-standing mathematical problem that had resisted human solution for eight decades, representing a significant milestone in AI's capacity for advanced formal reasoning. The breakthrough highlights how AI systems are increasingly able to tackle problems in domains that have historically required deep human expertise and creativity. Ars Technica provides a clearer breakdown of the solution than OpenAI's own announcement.
Read more →
Ars TechnicaHackers discovered a vulnerability in Meta's AI support chatbot that allowed them to take over high-value Instagram accounts by tricking the bot into changing account email addresses and resetting passwords. Stolen handles — often worth significant sums — were resold before Meta identified and patched the exploit. The incident raises serious concerns about the security risks introduced when AI systems are given the ability to perform account-level actions.
Read more →
MIT Technology ReviewChinese regulators have approved the world's first invasive brain-computer interface chip for clinical use, with an early patient — paralyzed from the neck down after a car accident — already demonstrating the ability to write using the implant. The development puts China in direct competition with Neuralink and other Western BCI efforts, and signals a new phase in the global race for neurotechnology dominance. Researchers say the approval opens the door to broader trials and more ambitious applications.
Read more →
Ars TechnicaGitHub has shifted Copilot to a usage-based pricing model, and developers are reacting with alarm as some report exhausting their entire monthly AI credit allotment in a single day of heavy use. The change exposes a growing tension between the promise of AI-powered productivity tools and their real-world cost at scale. The backlash highlights how pricing model design will be a critical battleground for enterprise AI adoption.
Read more →
Ars TechnicaNVIDIA has announced RTX Spark, an Arm-based processor combining a high-end RTX GPU with unified memory, targeting Windows laptops and mini desktop PCs. The move positions NVIDIA as a direct competitor in the consumer PC chip space, challenging both Qualcomm's Arm dominance on Windows and AMD's integrated graphics offerings. Partners including Microsoft, Dell, and HP are already building devices around the new platform.
Read more →
The VergeGoogle's new Gemini Spark agent, billed as a "24/7" autonomous AI assistant capable of handling tasks on a user's behalf, delivered genuinely impressive results in hands-on testing by The Verge. However, reviewers flagged significant concerns about the financial cost of sustained use and the privacy implications of granting an AI agent such broad access to personal data and workflows. The product represents Google's most ambitious push yet into agentic AI for consumers.
Read more →
The VergeMicrosoft is heading into its Build developer conference in San Francisco with plans to unveil new AI models and significant Windows improvements, in what observers are calling one of the most pivotal Build events in the company's history. The conference comes as Microsoft continues to restructure its entire business around AI, making developer buy-in more critical than ever. New Windows developer mode features and deeper Copilot integrations are among the expected announcements.
Read more →
TechCrunchWindBorne, an AI weather startup operating a global fleet of roughly 400 atmospheric balloons, is producing forecasts that are beating those of established government agencies in accuracy benchmarks. The company's edge comes from combining proprietary sensor data collected by its balloon network with advanced ML models trained on that unique dataset. The results suggest that AI-native approaches to data collection and modeling can outperform legacy institutional infrastructure.
Read more →
Ars TechnicaIntel is positioning its forthcoming Crescent Island AI accelerator as a more cost-efficient and thermally manageable alternative to offerings from NVIDIA and AMD, using LPDDR5 memory and an air-cooled design. The claims come as Intel attempts to claw back relevance in the AI chip market after ceding significant ground to its rivals. If the performance-per-watt and pricing claims hold up, Crescent Island could appeal to enterprises looking to reduce data center operating costs.
Read more →
NVIDIA BlogNVIDIA has unveiled Cosmos 3, described as the first open omni-model designed specifically for physical AI reasoning and action, enabling robots and autonomous systems to reason about the world before acting in it. The model is positioned as a foundational layer for next-generation robotics and autonomous machine applications. Its open release could accelerate development across the physical AI ecosystem in the same way open LLMs have accelerated software AI.
Read more →
NVIDIA BlogNVIDIA announced JetPack 7.2 and NemoClaw support for its Jetson edge AI platform at COMPUTEX, bringing agentic AI capabilities, CUDA 13 support, and Multi-Instance GPU functionality to embedded and robotics deployments. The update also delivers meaningful performance gains on the Jetson AGX Orin 32GB module. The release signals NVIDIA's push to bring the same agentic AI paradigms dominating cloud computing into physical, real-world devices.
Read more →
NVIDIA BlogBanks and financial institutions are moving away from fragmented, task-specific AI models toward unified transaction foundation models that can build a holistic understanding of customer financial behavior across siloed systems. The shift mirrors the broader industry trend from narrow AI tools to general-purpose foundation models, now applied to the specific demands of financial data at scale. NVIDIA argues this convergence will unlock new capabilities in fraud detection, credit risk, and personalized financial services.
Read more →
TechCrunchDuckDuckGo has released new browser extensions for Chrome and Firefox that make it easier for users to default to its AI-free search experience, capitalizing on a wave of users seeking alternatives to AI-infused search results. The company reports booming traffic as a growing segment of users express frustration or distrust with AI-generated answers in mainstream search engines. The move reflects a meaningful market opportunity in privacy-first, AI-optional web tools.
Read more →
Ars TechnicaGeneral Motors is deploying AI and machine learning across its engineering workflows — including computational fluid dynamics, finite element analysis, and digital twin simulations — to reduce tasks that once took 15 hours down to under a minute. The dramatic compression in simulation time is accelerating vehicle development cycles and enabling engineers to explore far more design iterations than was previously practical. GM's approach offers a concrete industrial case study for how AI is transforming traditional manufacturing R&D.
Read more →No stories from this source today.