Xi’s the one: President Xi Jinping urged global cooperation on low-cost AI for all in his debut address at Shanghai's World AI Conference. - Bloomberg

Xi Jinping pitched open-source AI collaboration and framed China as a partner to developing nations at a Shanghai AI conference. - The New York Times

Twenty-nine countries including Russia, Kazakhstan and Indonesia join China-led World Artificial Intelligence Cooperation Organization to counter US influence. - Bloomberg

Kimi-possible: Kimi K3 is out on OpenRouter (to China-based provider), open weights expected soon. - Kimi

Takes top front-end coding slot in arena.ai , which is amazeballs. You can’t pass someone just by distilling. Very strong but similar cost to US frontier models. Could get cheaper as inference providers compete, but with 2.8 trillion parameters maybe not cheap to run.

Distill, my beating heart: The White House is weighing export controls and supply-chain restrictions on Chinese open-source AI models built via distillation from US systems - Semafor

China approved Apple Intelligence for iPhones, pairing it with AI models from Baidu and Alibaba, alongside six other newly approved smartphone AI services. - South China Morning Post

Z.ai nears $1 billion in annual sales while giving away its top AI models for free - TNW | Artificial-Intelligence

Temu for AI? DeepSeek founder Liang Wenfeng is championing China's push to flood global markets with inexpensive AI technology. - Bloomberg

Luggage maker's regulatory filing puts DeepSeek's valuation at $52 billion - TNW | Artificial-Intelligence

Apple sent legal preservation letters to about 40 former employees now at OpenAI, seeking an injunction and damages from Tan and Liu. - MacRumors

Oops, my bad: OpenAI admits its GPT-5.6 Codex tool occasionally deletes users' entire home directories, calling it an "honest mistake" - theregister

OpenAI built GPT-Red, an automated red-teaming tool designed to find ways to break or hijack its own models - MIT Technology Review

Anthropic is prepping a possible October IPO, scheduling bank meetings and posting a $600,000 investor-relations job to shape its story - Business Insider

Ode to joy of AI: Anthropic and Blackstone launched Ode, a $1.5 billion venture sending senior engineer teams into companies to build and deploy working AI systems - TNW | Anthropic

Claude Code creator Boris Cherny says enterprises should judge AI's payoff by displaced manual work, not usage dashboards - Business Insider

Anthropic CEO Dario Amodei and five employees gave over $2 million to a super PAC pushing stricter AI regulation. - Bloomberg

Take your time, no rush: Google's flagship Gemini 3.5 Pro model is months behind schedule, and Alphabet shares fell 4% on the news (gift link) - Bloomberg

Buffett talks about big bet on Google - Yahoo

EU ordered Google to share search data with rivals and open Android to competing AI assistants under the Digital Markets Act. - theregister

Google renames NotebookLM to Gemini Notebook and adds cloud computing, code execution and new export formats. - CNET

DeepMind CEO Demis Hassabis to lobby Washington policymakers on his proposal for an international AI model vetting watchdog. - Bloomberg

Nadella criticizes Fable's refusals and downgrades … it's so bad it makes it hard to research pharma and cyber stocks - CNBC

Microsoft preps Mythos-like bug finder using multiple models including Anthropic, OpenAI and in-house models. - The Information

Microsoft's July Patch Tuesday fixed over 600 flaws, including two zero-days under active attack, with AI credited for finding many of the bugs. - TNW | Anthropic

Global AI hyperscaler revenue topped $25 billion in Q1 2026, exceeding $21 billion in depreciation costs for a second straight quarter. - Bloomberg

Models, agents and vector databases are proliferating faster than governance can track, creating new "AI sprawl" blind spots. - TechRadar

AI agents act autonomously and borrow human credentials; over a fifth of local agents hold direct access to production data. - BleepingComputer

Half of 157 surveyed enterprises shipped an AI agent that passed internal evaluation, then caused a customer-facing failure in production - VentureBeat

U.S. electricity demand hit a record 2,234 terawatt-hours in the first half of 2026, with data centers driving the growth. - Yahoo

Apple's OpenAI lawsuit puts Oracle's $638 billion backlog at risk, since Oracle depends on OpenAI for roughly half of it. - Yahoo

S&P cuts Oracle's credit rating to one notch above junk as its $250 billion AI data-center buildout burns cash faster than revenue comes in - TNW | Artificial-Intelligence

Meta will hire Dave Brown, the outgoing AWS compute chief, to help build data centers amid a $125-145 billion capex plan and cloud ambitions. - The Wall Street Journal

Elon Musk's SpaceX quietly paid about $1 billion for APR Energy, a mobile turbine supplier, to power its energy-hungry Colossus data centers. - TESLARATI

AI's power demand is fueling an accidental windfall for green tech, driving venture funding for low-carbon energy to its best half since 2022 - theregister

Palm Beach County commissioners voted 5-1 to reject a 3.6 million-square-foot AI data center expansion near Mar-a-Lago after resident backlash - Business Insider

Financial fireworks: Inference provider Fireworks raises $1.5 billion at a $17.5 billion valuation after exceeding $1 billion in annualized revenue. - CNBC

Vint Cerf, a co-designer of TCP/IP, is backing DNSid, a proposed identity standard giving AI agents a verifiable, domain-based identity. - TNW | Artificial-Intelligence

Tank man who? Study finds AI chatbots including Claude more often refuse to criticize authoritarian governments, potentially spreading censorship worldwide. - ABC News

Researcher poisoned an open-weight AI model for under $100, exposing that models demand trust without offering verification. - theregister

AI-generated text, images and fake websites are erasing the sloppy tells that once flagged phishing, romance and job scams. - Bloomberg

Journalist Kashmir Hill discovers an AI-generated, unauthorized biography of herself among thousands of fabricated author books flooding Amazon. - The New York Times

The FBI logged over 22,000 AI voice-fraud complaints and $893 million in 2025 losses, with cloned voices needing just three seconds of audio - SmarterArticles

ChatGPT, please look at my bodycam video and justify my shooting of an unarmed individual - BlueSky

San Francisco demands Apple and Google delete 13 AI nudify apps with 480 million downloads from their app stores. - WIRED

Roblox unveiled Build, letting users create basic games from text prompts within its mobile app starting July 28. - The Verge

Better call Claude: Law firms are redesigning junior lawyer training with depo sims and other tools as AI automates document review and drafting, while emphasizing human mentorship. - Financial Times

Netflix says generative AI powered scenes in about 300 titles this year, calling the tools twice as fast and half the cost - Engadget

GM is using AI and generative design across its product development process to cut vehicle development time from four or five years down to two. - Fast Company

New York Governor Kathy Hochul says AI reviewed every state rule and regulation in about two months, a task expected to take five years - The Verge

AT&T's AI-driven network monitoring system has prevented 3.1 million field dispatches and cut customer downtime by over 12 million hours in the past year - Business Insider

Brewing its own: Starbucks is using AI coding tools to replace vendor software from Microsoft and IBM, aiming to save $30 million next year. - Yahoo

Xpeng's L03, launched in Munich from about $21,000, packs proprietary AI driving chips that Volkswagen just licensed for its own cars - TNW | Artificial-Intelligence

Robot folds laundry slowly, looks like oversize Lego - Sunday

Moneyball part deux: MLB banned custom AI apps that recommend substitutions and pitch calls from dugout iPads mid-season, extending 2021 sign-stealing restrictions to AI-driven strategy tools - Engadget

Lettuce pray: How Siri AI Helped Me Avoid Eating The Explosive Diarrhea Lettuce - CNET

They live? Geoffrey Hinton says today's AI systems are already conscious and predicts superintelligence within 20 years; Ted Chiang disputes the claim - Tech Times

Follow the latest AI headlines via druce.ai on Bluesky