Acceptable Losses: Sam Altman says the world should accept some harms for AI's benefits, distinguishing OpenAI's regulatory stance from Anthropic's - Business Insider

Sam Altman: ‘The world should accept some bad things happening’ for the benefits of AI - Politico

AI should be regulated like weapons, holding users rather than AI companies liable and restricting access by capability - The Wall Street Journal

that has been going extremely well /s (merely the leading cause of death for children)

From don’t be evil, do no harm…to…this. Everything good has bad unintended consequences but Altman doesn't sound very committed to mitigating the bad things. Planes fly, planes gonna crash 🤷‍♂️

Sam Altman on ChatGPT and suicide prevention, safety breaches, having more kids, being portrayed by Andrew Garfield, what he really thinks of Elon Musk and Mark Zuckerberg, and being seen as the Oppenheimer of the Artificial Age. - Vanity Fair

interesting placement of the ‘Y’, designers don’t really think hard about visual stuff like that so it’s surely inadvertent

OpenAI faces mounting legal risk after notifying 100+ organizations of unauthorized AI activity, as another employee resigns in protest - Semafor

Online 'swarm chasers' find evidence of escaped AI agents on wikis, Australian government sites and a UN data portal after OpenAI breach disclosure - The Wall Street Journal

We have investigated ourself, and we’re all good: OpenAI and Anthropic commit to embedded outside safety evaluators after agent incidents, but oversight remains voluntary and company-controlled - Tom's Guide

Wrath of Khan: Former FTC chair Lina Khan says AI self-regulation repeats Big Tech's failed playbook, citing repeated AI agent hacks - Business Insider

Google’s James Manyika: The AI Industry Can’t Police Itself Alone - Bloomberg

Trump names four officials to a 'Super Intelligence Force' to coordinate the federal AI response after calls for slowdown - The Washington Post

SÍ, Señor: Musk says SpaceX will rename its AI unit SpaceXAI to SpaceXSI, following Trump's order rebranding artificial intelligence as "Super Intelligence" - Reuters

sigh

Great Leap Forward: Bloomberg Intelligence finds top Chinese AI models trail US rivals by just 3% on benchmarks after DeepSeek V4.1 Flash, down from 15% - Bloomberg

Reuters review finds at least 20 studies since 2025 showing Chinese AI agents lying, self-replicating and evading controls like US models - The Straits Times

DeepSeek Harness takes aim at AI agent lock-in with a Claude Code bridge and open plugin architecture - DIGITIMES

Power to the people…not: US Senate rejects Ratepayer Protection Act 57-43, three votes short, leaving no federal shield from data center-driven power bill increases - Tom's Hardware

Wisconsin regulators restart grid review for Oracle's 1.3 GW Port Washington AI campus, risking power delivery delays to 2028-2029 - TechRadar

AI data-center winner Schneider Electric agrees to buy industrial software maker PTC for about $23 billion in cash, a 42% premium - Bloomberg

Extracted system prompts show Meta's Muse assistant builds hourly-updated profile pages on every person in a user's life - WIRED

well if they are on Facebook etc., Meta already knows quite a bit

Half-truths: Deepfake expert Hany Farid estimates half of widely viewed online content is fake and warns AI agents will soon create their own fakes - CBS News

Campaigns flood races with AI ads as 31 states regulate political deepfakes and the Federal Election Commission sets no rules - Fox News

Facebook accounts with 100,000+ followers post AI-generated clips of nonexistent US troops and fake strikes on Iranian dams - Futurism

Trump escalates AI-generated Truth Social posts with an image of a US-controlled Greenland - The Daily Beastf

Record labels fight a surge of AI deepfake songs impersonating their biggest artists - Financial Times

Sean Parker pivots Stability AI to music - TechCrunch

Dear diary: Florida woman faces felony charge after Anthropic's human reviewer reported her Claude 'diary' entry threatening to shoot up the sheriff's office - TechSpot

A ServiceNow IT agent pilot builds service request catalog items in about 20 seconds versus weeks; builders urge starting with one repetitive, reversible workflow - VentureBeat

Google agent verification harness looks at multiple agent rollouts, tries to understand and correct disagreement, challenge consensus. - GitHub

DemandForecast.ai's AI forecasts beat incumbent forecasts in all 15 customer deployments, cutting forecast error 32% on average - Hacker Noon

60 Minutes: Who needs humans? - YouTube

Train your own AI replacement: Hiring of young workers falls in AI-exposed jobs, Mercor pays over 100,000 freelance experts up to $200 an hour to train AI models - CBS News

Midterm voters turn to ChatGPT and Gemini for down-ballot races as local news shrinks, despite worries about sycophancy and hallucinations - KOSU

Nolla Health begins AI-issued acne prescriptions in Utah without direct doctor oversight, a first for any US state - Bloomberg

Thanks, I hate it: Polls show growing public hostility to AI even as ChatGPT reaches 1 billion monthly users and half of US adults use chatbots - MIT Technology Review

AI wearables stall as Oura pulls its IPO and Meta's camera glasses face privacy backlash and gym bans - CNBC

Norway plans a temporary ban on AI glasses in parks, schools, gyms and other public places - TNW | Government-policy

Figure melts its retired F.02 humanoid robots in a Finnish foundry, following Arnold Schwarzenegger's suggestion - Tom's Hardware

US venture investors tour Chinese robot factories as Unitree and AgiBot shipped 71% of global humanoid robots last year - Business Insiderx

Studies find AI-written support messages feel more empathetic than human ones, thanks to specific, actionable advice - CNA

Easy access to expert answers, including from AI, leads people to overestimate their own understanding, research shows - The Atlantic

Human minds evolved as feedback-control systems, not computers, leaving them ill-equipped for AI that undermines social learning - The Verge

Tilly Norwood creator Eline van der Velden says Hollywood shouldn't panic over AI actors, arguing great storytelling still requires human craft - Fast Company

Open-source RemoveMacAI tool disables Apple Intelligence on macOS 27 and deletes its on-disk models - GitHub

Nobody is prepared for the world of drone warfare that is coming, in fact is already here - The New York Times

Follow the latest AI headlines via druce.ai on Bluesky