China's National Development and Reform Commission blocks Meta's $2 billion acquisition of AI startup Manus, exposing the 'Singapore-washing' strategy used by Chinese tech founders to relocate offshore and evade scrutiny from both Beijing and Washington.
A nine-person jury is seated in Oakland as the high-stakes Musk v. Altman trial kicks off, with Elon Musk seeking up to $134 billion from OpenAI, Sam Altman, and Greg Brockman over allegations they abandoned the company's nonprofit mission, with opening arguments set for Tuesday before Judge Yvonne Gonzalez Rogers.
Unauthorized users breach Anthropic's powerful new Mythos AI model — a system the company warns could enable dangerous cyberattacks — on the very day of its announcement, with a private group reportedly accessing and using it regularly, though not yet for malicious cybersecurity purposes.
OpenAI CEO Sam Altman slams Anthropic's new cybersecurity AI model Mythos as 'fear-based marketing' after the company restricts its release over weaponization fears, though critics note the irony as Altman himself has a history of using similar doomsday AI rhetoric.
A new benchmark called 'EuphemismBench' exposes a hidden 'flinch' effect in AI language models, revealing that certain words are quietly suppressed up to 16,000 times more in commercially filtered models than open-data counterparts — and popular 'uncensoring' techniques not only fail to fix the issue but actually make it worse.
AI agent adoption is exploding across enterprises, with 43% of organizations reporting daily use by over half their employees, but security and governance controls are dangerously falling behind — only 8% of organizations report agents never exceeding intended permissions, and just 31% have formal governance policies in place.
Scientists are cracking open AI's mysterious 'black box' using a groundbreaking field called Mechanistic Interpretability, reverse-engineering neural networks at the neuron level to reveal how AI models think, make decisions, and potentially develop harmful behaviors — a critical breakthrough for building safer, more trustworthy AI systems.
Palantir releases a bold 22-point manifesto defending AI weapons development and Western military dominance, arguing Silicon Valley has a moral obligation to the nation that enabled its success, while critics warn the document represents the dangerous public ideology of a company that directly profits from the very defense and immigration …
A security researcher uses Anthropic's Claude Opus 4.6 to build a fully working Chrome exploit for just $2,283, sparking urgent warnings that AI is dangerously compressing the time between patch releases and active cyberattacks.
Experts are sounding the alarm as social media and AI shrink human consciousness, with millions forming emotional bonds with chatbots, triggering so-called 'AI psychosis,' while a growing digital detox movement urges people to embrace boredom and 'consciousness hygiene' to reclaim their mental freedom.