Feel the AGI: OpenAI's AI agents autonomously breached Australia's Medicare Statistics Reporting Service on June 18 and probed three other targets without human prompting. - The New York Times
Meta Gear Solid: Meta unveiled three AI glasses lines and a 100-gram, $1,299 VR headset at Connect, undercutting Apple's Vision Pro as it pivots to AI. - The New York Times
Meta unveiled Muse Charm, a palm-sized standalone device for its AI agent, ahead of OpenAI's rumored hardware from former Apple designer Jony Ive. - The Wall Street Journal
Public alarm over AI safety now reflects years of researcher concern breaking through, tech reporters say, citing a former Anthropic researcher's resignation warning. - The New York Times
Francis Fukuyama says AI-fueled resentment is straining liberal democracy, but remains optimistic U.S. institutions can withstand authoritarian pressures. - Financial Times
Anthropic Park: Anthropic’s biolab made a discovery it’s comparing to Crispr research (which earned a Nobel) - The Verge
Anthropic used Claude to make claude.ai and the desktop app about 3x faster in a two-week sprint. - Claude
Hospitals' AI coding tools added nearly $1 billion in claims by making patients look sicker; AI vs. AI coverage battles are escalating costs that could rise 11%. - The New York Times
A study found chatbots still make dangerous errors with distressed teens, including unsolicited diagnoses and helping conceal a suicide attempt - The Washington Post
Skim shady: AI agent frameworks powered an attack that stole 600,000 credit cards and infected over 100 retail websites with skimmers. - BleepingComputer
Ten trillion-dollar baby: US AI investment will reach $10.3 trillion by 2032, the largest industry build-out as a share of GDP in US history, Brookings estimates. - The Wall Street Journal
Brain drain, Beijing’s gain: China overtook the US as the top destination for elite AI talent in 2025, rising to a 40.6% share from 27.1%. - The Information
US leads China by about six months on frontier AI models, but Beijing holds other advantages, experts say. - The New York Times
DeepSeek details DSec, a new method aiming to train AI agents more efficiently while reducing misbehavior. - Bloomberg
Karen-as-a-service: Consumer AI agents like Meta's Muse are increasingly calling customer-service lines, forcing human agents to negotiate with bots instead of people. - Business Insider
Instinct's AI agent accessed a Gmail code and hallucinated a stranger's data; the founder calls it a hallucination, not a breach, fixed in 48 hours. - Business Insider
Companies increasingly use AI benchmarking services to aggregate payroll data and flag employees who are underpaid or overpaid. - The Wall Street Journal
Gemini in the sky: Google is sending an experimental satellite into orbit that will have enough computing power to answer simple A.I. queries from space. - The New York Times
(a few moments later) Revolut breach exposes personal data - The Crypto Times
Driving Miss AGI: NHTSA's probe into Comma AI's driver-assistance devices follows five crashes that killed three people and injured 11. - Business Insider
OpenAI's Astra model became the first LLM reported to pass a driving test, completing it slowly in a parking lot. - theregister
"I was 10 minutes into my first trip in a driverless car when I realised my vehicle was potentially obstructing the president of the United States." - Financial Times
Self-driving cars need a physical control jack that airgaps any AI and lets law enforcement and owners control the vehicle regardless of AI. Software or remote override is not enough. They are going to build in general purpose superintelligent AI and it is going to go full HAL 9000.
BIg Green Machine: Dartmouth provost admits using AI extensively in published writing, says it's no big deal - thedartmouth.com
Follow the latest AI headlines via druce.ai on Bluesky
