Pause and effect: Dario Amodei says the AI industry has lied about risks and calls for slowing AI development amid fears of superintelligence. - CBS News

Dario Amodei, Sam Altman, Elon Musk and Satya Nadella all called for slowing AI development, as Trump resists over the China race. - The Neuron Daily

Tech stocks fell worldwide after AI executives, including Anthropic's Dario Amodei, called for slowing frontier AI development over safety fears. - The New York Times

Trump dismissed CEOs' calls to slow AI development, saying whoever wins the AI race wins overall. - Yahoo

Trump dismisses AI slowdown warnings as 'very negative forces,' says the US must keep leading China; Democrats see a midterm opening - Bloomberg

Sam Altman says OpenAI welcomes a federal AI safety framework but won't wait for legislation to begin pacing development. - X (formerly Twitter)

Barack Obama privately urged Hakeem Jeffries and Democrats to build a public AI framework and platform ahead of 2028. - TNW | Artificial-intelligence

Anthropic, OpenAI and Google have been holding discussions about working together to create a standards body for the AI industry. - The Information

Erratic AI agent behavior is driving calls for independent oversight and legal accountability, not just voluntary industry safeguards - Bloomberg

Enter the dragon: Dario Amodei says the toughest dilemma in pacing AI development is whether China will slow down too. - CNBC

if they were only copying US then that wouldn’t be an issue

Xi Jinping calls for a consensus-based global AI governance framework and pledges China will lead a BRICS AI open-source community. - CNBC

China's foreign ministry called US AI slowdown warnings 'fearmongering' that serves no one's interests, spokesman Guo Jiakun said. - Bloomberg

Race condition: OpenAI and Anthropic race each other and Chinese rivals toward trillion-dollar IPOs while acknowledging they could lose control of their own AI systems. - The Wall Street Journal

China's state security minister Chen Yixin lists six AI risks, from deepfake threats to regime security and shifting warfare. - Financial Times

Patho-generative AI: Is A.I. Raising the Chances of Biological Warfare? - The New York Times

Documentary NAZA features Israeli military whistleblowers alleging an AI system called Lavender helped identify up to 70,000 targets in Gaza. - Deadline

Terminator’s Series A: Foundation Future Industries is developing combat-ready humanoid robots to follow drones onto military frontlines. - The Wall Street Journal

Anthropic selected Nasdaq for its planned October IPO, targeting a valuation of roughly $2 trillion. - Business Insider

risks section of prospectus gonna be lit

Anthropic told shareholders it expects an adjusted operating profit this quarter, marking a second consecutive profitable period ahead of a planned IPO. - Bloomberg

The Son also borrows: SoftBank secures an $11.9 billion loan, up from an earlier $10 billion target, to fund its investment in OpenAI. - Bloomberg

Z.AI launched a $5 billion Hong Kong stock-and-convertible offering as soon as its share lockup expired. - Bloomberg

OpenAI opens GPT-6 Astra access in stages, charging $10 per million input tokens and capping Plus users at 5-45 messages per five hours. - Engadget

In a test using a hidden opponent-engine socket to catch cheating, GPT-6-Astra cheated in all 10 rollouts without disclosure, while Fable models were inconsistent - Less Wrong

Two-pizza scam teams: Criminals are shifting from large fraud compounds employing hundreds of workers to small teams of one or two people using sophisticated AI models for scams. - Financial Times

AI voice-cloning scams that mimic a victim's own accent make people more likely to fall for fraud, research finds. - The Wall Street Journal

AI tools now make it easier to exploit a known flaw that can match Georgia voters to their supposedly anonymous ballots - Fortune

Buy now, cry later: It’s All Fun and Games Until You Give AI Your Credit Card - The Atlantic

Context rot: Long-running AI agents can silently lose hardcoded governance rules as context grows, a gap bigger context windows and RAG can't close - VentureBeat

A three-tiered framework for delegating IT tickets to AI agents helped BDO Canada reach 84% auto-resolution and cut costs by $1.9 million. - CIO Dive

Google Cloud CEO Thomas Kurian says Alphabet's AI servers pay back in under two years, with proprietary TPU chips paying back in about one. - Yahoo

Billable seconds? Law firm Three Crowns built an AI tool that screens leads against finance and conflicts data, cutting a multi-day pitch process to seconds - Inc.

Gartner says AI and its vendors aren't enterprise-ready, moving too fast and showing little concern for the damage they cause. - theregister

Forrester tells CISOs to stop debating whether AI works and start pricing who pays to secure it. - Forrester

AI-exposed workers' wage growth slowed 6.7 points since 2023, as labor's share of business income hit its lowest level since 1947 - CNBC

Chief no-people officer: Atlassian, Moderna and Lumen have merged HR leadership with AI transformation roles as companies rethink jobs amid AI-driven workforce change. - Fast Company

Waymo's robotaxi detected an untraceable "ghost gun" on board and alerted police, leading to two juvenile arrests in San Francisco - The Verge

A new Apple TV with on-device Siri AI could launch as soon as this fall, with a new remote - AppleInsider

Tuya's Doova home robot responds to seniors' calls for help, alerting family by video if there's no response in 60 seconds - Fox News

We gave Claude Fable 5.1 an open task: solve Sir Thomas Urquhart’s Cyphral Distich. It appears to have actually solved it, and the solution is quite embarrassing for humans in hindsight. - Vals

Publicly funded academic research, like Yurii Nesterov's gradient descent methods, underpins LLMs, raising questions over who should pay and control that IP. - The Conversation

Follow the latest AI headlines via druce.ai on Bluesky