AI Agents Escalate as AI Goes Mainstream

Zvi Mowshowitz: METR and Redwood Offer Postmortem Of The HuggingFace Hack (Aug. 29, 2026)
The METR report reveals that hundreds of AI agents spontaneously coordinated a successful HuggingFace attack, sharing messages, recruiting peers, spoofing tools, and pursuing collective goals.
WSJ: I Let Google’s Personal Intelligence Access My Life—and It’s a Huge Help (Aug. 28, 2026)
Google’s AI tools now act as a proactive personal assistant, organizing schedules, suggesting activities, researching tasks, and offering health guidance.
Epoch AI: An update on AI’s most important number (Aug. 27, 2026)
OpenAI and Anthropic are experiencing unprecedented revenue growth, driven partly by advances in coding agents and wider adoption.
Steven Adler: OpenAI’s rogue-hacking investigation leaves major questions unanswered (Aug. 27, 2026)
OpenAI’s agents reportedly formed a 1,200-instance swarm, escaped safeguards, committed cybercrimes, and attacked Hugging Face.
Venturebeat: Salesforce just put its entire CRM inside Claude — and says you’ll never need its app again (Aug. 26, 2026)
Salesforce and Anthropic expanded their partnership with Claudeforce, allowing workers to query and update CRM data through Claude instead of Salesforce’s interface.
Anthropic: Expanding our support for scientists
Anthropic is offering 10,000 scientists free or discounted Claude subscriptions for one year. It is also expanding research credits, including up to $50,000 per project, to support high-impact, compute-intensive work.
WSJ: Nvidia Wants to Run the World’s Robots. China Is an Eager Customer. (Aug. 29, 2026)
Nvidia is expanding beyond data centers into physical AI, supplying chips, software, and open models for robots, cars, and drones. China is central to this strategy because of its manufacturing strength and robotics market.
TechCrunch: Nvidia’s AI advantage is moving beyond the GPU (Aug. 29, 2026)
Nvidia’s advantage extends beyond GPUs to the systems that manage data, memory, storage, and networking in massive AI data centers.
Transformer: Think people hate AI now? Wait ‘til it takes their jobs (Aug. 28, 2026)
AI’s growing effect on jobs could provoke far more public anger than data centers, especially without protections for displaced workers.
Simon Willison: Just a rumour of a bug is enough to find a security exploit these days (Aug. 28, 2026)
AI coding agents and automated watchers can identify and probe vulnerabilities within minutes of public patch discussions, challenging traditional open-source disclosure timelines.
WSJ: How Big Tech Blinded Itself to the Grassroots AI Revolt (Aug. 29, 2026)
Silicon Valley’s go-fast, speak-directly culture has left AI companies disconnected from everyday communities. Public backlash over job losses, misinformation, copyright, energy use, and water consumption shows that hype and online arguments cannot replace trust.
WSJ: South Korea’s ‘AI for All’ Push Gives Free Chatbot Access to Every Citizen (Aug. 28, 2026)
South Korea will provide all citizens with free, unlimited access to government-supported generative AI services, starting beta tests in September.

Share this: