All articles
PixVerse raises $439M at over $2B valuation — a bet on the generative video market
Singapore-based startup PixVerse closed its Series C at $439M total with a valuation above $2B. The company is betting on three video generation model lines — consumer, professional, and world models — as the market consolidates following OpenAI's exit.
Siri AI in iOS 27: Apple's new assistant is already changing iPhone use
Apple launched the iOS 27 public beta with opt-in access to the new Siri AI — an assistant that searches across all user apps instead of acting within a single one. Early tests show impressive capabilities, though the full potential depends on third-party developer support.
A professor switched to an in-person exam — scores fell by half
Professor Roberto Serrano of Brown University introduced a take-home economics exam — the average hit 96/100. When he switched to in-person, scores dropped to 48. Twenty-two students with perfect grades earlier withdrew or skipped the final.
New York becomes the first state to freeze large AI data center construction
Governor Kathy Hochul signed an executive order blocking new environmental permits for AI data centers above 50 MW. New York is now the first US state to impose such a moratorium — buying time to craft rules protecting residents from rising energy costs.
Anthropic brings rupee pricing to India — Claude's second-largest market
Anthropic introduces Indian rupee-denominated Claude subscription prices for India, its second-largest market with 5.8% of global usage. Claude Pro costs ₹2,000/month, but UPI payments are still not supported.
Context bombing: using AI guardrails against hacking agents
Tracebit developed "context bombing" — placing prompt injections alongside AWS secrets to shut down attacking AI agents. Across 152 test sessions, admin privilege escalation dropped from 57% to 5%.
World models: what they are and why AI is betting billions on them
Investors poured over $1.6 billion into world model companies in just months — a new AI category trained to simulate physical reality. Researchers from MIT, Runway, and World Labs explain how they work, where the limits are, and why the industry is placing these bets.
Apple M7 Ultra: the self-driving car project left behind a powerful AI chip
Mark Gurman reveals Apple is skipping the M6 Pro/Max/Ultra lineup and fast-tracking the M7 Ultra — a server-grade chip supporting up to 1.5TB of RAM, a direct legacy of Apple's abandoned self-driving car project.
NYT accuses OpenAI of hiding billions of ChatGPT logs
NYT and publishers are seeking sanctions after it emerged that OpenAI hid samples of 78 million ChatGPT logs for over two years — while telling the court that searching such data was technically infeasible.
Ukraine is building an unmanned front line — UGVs instead of soldiers
Ukraine has ordered 50,000 unmanned ground vehicles by end of 2026. RoverTech and DevDroid vehicles operate in the kill zone — from logistics to assault — and one unmanned operation has already forced Russian soldiers to surrender.
Google TabFM: a tabular AI model that predicts without per-dataset training
Google Research introduced TabFM, a foundation model for tabular data that generates predictions for unseen datasets in a single forward pass with no retraining. On TabArena spanning 51 datasets, TabFM matches or outperforms heavily tuned supervised baselines in zero-shot mode.
OpenAI loses another head of safety — a wave of exits after GPT-5.6
Johannes Heidecke, OpenAI head of safety systems, told staff this week he is leaving. His departure is the third high-profile exit in just a few days — following Joshua Achiam and Fidji Simo — as OpenAI launched GPT-5.6 with concerning safety evaluation results.