Monitor
The NSA is accessing Anthropic's powerful Mythos Preview model for security tasks, even after Trump admin labeled Anthropic a risk and feud with Pentagon. About 40 orgs have preview access to the cyber-capable model announced early April.
Palantir published a statement criticizing DEI initiatives and 'regressive' corporate cultures, fueling debates on tech industry values amid AI talent wars.
Anthropic's head cautions that Mythos, the latest Claude family addition, poses significant risks and won't be broadly released. It excels at finding software vulnerabilities but raises doomsday concerns.
AI coding startup Cursor is negotiating a $2 billion funding round led by a16z with Nvidia and Thrive Capital participating, pushing its valuation above $50 billion pre-money. The round reflects explosive demand for AI developer tools amid agentic coding hype.
Frontend cloud platform Vercel reported a breach affecting a limited number of customers, underscoring growing cybersecurity threats to AI infrastructure.
Google parent Alphabet's shares tumbled amid investor concerns over ballooning AI infrastructure spending, despite strong cloud growth. Analysts debate if the capex frenzy for TPUs and datacenters justifies the valuation hit.
The Trump administration seeks federal preemption on AI rules to prevent a patchwork of state laws, but Utah lawmakers resist, prioritizing local control on emerging tech risks.
Renewable energy startup Panthalassa unveils sea-based data centers using ocean waves for power, targeting AI's massive energy demands and reducing carbon emissions from land-based facilities.
Alphabet's Google is discussing with Marvell Technology the development of two chips: a memory processing unit to complement its TPUs and a new TPU optimized for AI models. The goal is to enhance efficiency in running AI inference, reducing dependence on Nvidia GPUs. Design finalization for the memory unit could happen next year.
A leading AI expert argues against unregulated AI development, advocating for targeted oversight to mitigate risks without stifling innovation. The piece comes amid intensifying global policy debates.
Ex-DeepMind and OpenAI team behind Recursive Superintelligence raises funds at $4B valuation, despite early-stage AI research focused on advanced capabilities.
A Utah Republican and former Google employee advocates for federal AI oversight to preempt patchwork state laws, amid Trump's broader deregulation agenda.
Early AI agent deployments like OpenClaw reveal persistent problems with efficiency and execution, challenging Nvidia CEO's bold predictions.
Nathan Benaich's newsletter highlights Q1 explosions: Anthropic rev from $14B to $19B, OpenAI at $840B valuation, signaling massive AI commercialization.
Anthropic touts Claude Opus 4.7 as possessing 'taste' alongside smarts and empathy, amid debates on anthropomorphizing AI capabilities. Benchmarks show gains but critics question subjective claims.
Alibaba unveils Happy Oyster, a new AI model specialized in generating 3D environments for gaming, signaling China's push into AI-driven content creation.
A team deploys Claude-powered AI to call thousands of stations for data collection, sharing tips on scaling AI agents for real-world tasks.
Bitcoin miner Iren is shifting to AI cloud services with massive GPU deployments, stock surging on pivot to high-demand AI compute.
Analyst Beth Kindig forecasts NVDA data center revenue doubling to $480B by 2028, driven by AI inference and CUDA ecosystem dominance.
TrendAI teams up with Anthropic to enhance AI security amid rising concerns over model vulnerabilities exposed by tests like Claude Mythos.
Honor Robotics' Lightning humanoid completed the 21km race in 50:26, beating the human record, as over 100 robots raced humans in a showcase of physical AI endurance and balance.
Honor's Lightning robot claimed victory in the 2026 Beijing E-Town Humanoid Robot Half-Marathon, finishing in 50 minutes 26 seconds ahead of humans; ~40% of entrants were autonomous, highlighting rapid advances in robot endurance and AI navigation.
Grok 4.3 beta (0.5T params) introduces PDF, slide, and spreadsheet creation for SuperGrok Heavy users; 1T version imminent per Elon Musk. No official benchmarks yet, highlights platform info gaps vs. competitors.
Salesforce unveils Headless 360, a comprehensive AI platform for decoupled frontends and personalized experiences, aiming to redefine enterprise AI deployment.