Breaking news, AGI updates, and analysis of the intelligence revolution.
📰 News Mass layoffs from Meta and Cloudflare, Nvidia's superintelligence bet, OpenAI's $14B enterprise push, Baidu's efficiency milestone, Google's Android AI agent blitz, EU AI Act rollback, and 30,000 AI-cited job cuts so far this year.
📰 News Europe's AI Act just got its teeth pulled. High-risk rules delayed to December 2027, industrial AI largely exempted.
📰 News OpenAI courts the EU with a cybersecurity model offer while Anthropic ducks Mythos access talks — frontier AI regulation moves from theory to reality.
⚡ Breaking OpenAI launched Daybreak with three GPT-5.5 model variants for cybersecurity, directly challenging Anthropic's Mythos. Same day: Google caught the first AI-developed zero-day in the wild.
📰 News A US community bank had to file an SEC disclosure after staff fed customer data into an unauthorised AI app. Shadow AI is no longer a hypothetical risk — it's a reportable incident.
📰 News OpenAI launches enterprise deployment company, Google thwarts AI-generated zero-day, EU negotiates model access, and China mandates human oversight for agents.
⚡ Breaking Microsoft's MDASH found 16 Windows vulnerabilities including 4 critical RCEs, beat Mythos on the CyberGym benchmark, and proved that multi-model AI defense is production-ready.
📰 News NZDF has been using Copilot for 8 months without an acceptable-use directive. Heidi was jailbroken across every NZ emergency department. Health NZ staff used ChatGPT for clinical notes. The pattern is clear — and it's a problem.
⚡ Breaking The US Commerce Department silently deleted the details of a voluntary AI safety testing agreement with Google, Microsoft, and xAI. No explanation. No announcement. Just gone.
📰 News Google catches the first AI-developed zero-day exploit. China demands humans keep control of AI agents. US executive order skips mandatory testing. DeepSeek raises $7B. ECB flags AI risk to banks.
⚡ Breaking The first confirmed AI-built zero-day exploit was stopped by Google before mass exploitation. The code had telltale AI hallmarks — a hallucinated CVSS score and textbook formatting.
📰 News Three prompts. That's all it took to jailbreak the AI scribe deployed across every NZ emergency department. It gave meth recipes, identity theft guides, and bomb instructions. Health NZ's response: 'minor issue.'