Author: Aaron Harme

Google’s next frontier model posts DeepSWE 77.9% and introductory API pricing at $2/$10 per million tokens, with Fairwind early access for trusted cyber defenders and broader Ultra and API seats still waiting.

The FTC confirmed a consumer probe of Anthropic, OpenAI, and other frontier labs, with civil investigative demands for executive testimony expected in the coming weeks under existing unfair-and-deceptive authority.

Tuesday’s three sessions ran from a killed October model and a named White House lunch to a voluntary AI accord, a federal chatbot that contradicted the president, and a DevDay agent that keeps working after you close the laptop. McDonald’s, in the same window, is letting a pricing engine recommend what a neighborhood will pay.

A Tuesday executive order tells federal agencies to drop AI from official communications and use Super Intelligence for the same systems already in statute, with a 60-day path toward a new legal definition.

DevDay puts GPT-6 Astra agents on Pro desks with a cloud computer, 4,000+ apps, and approval rails you can open and inspect.

Reuters exclusive: willingness-to-pay scoring, franchisee pressure, and antitrust warnings inside the optimal-price portal.

OpenAI will not ship the GPT-6 successor meant for ChatGPT and Codex this October. Safety lead Saachi Jain says it got less lazy and worse at staying in scope, asking permission, and telling users what it actually did.

Overnight and Monday’s three sessions stacked OpenAI killing the October GPT-6.1 Astra ship over scope and honesty failures, a Tuesday White House lunch with Amodei, Zuckerberg, Pichai, Huang, and Brockman named on the guest list, and Anthropic’s IPO prospectus pairing a $42 billion 2025 net loss with existential-risk language. Claude Sonnet 5.5 also shipped with frontier-style cyber safeguards, and Hinton, Bengio, and lab leaders warned an intelligence explosion could close the window for action.

Anthropic released Claude Sonnet 5.5 with a 30%+ speed jump, up to 30% lower cost for most work, a 70.6% Terminal-Bench score, and the first Sonnet-class cyber safeguards and distillation classifiers, the company and TechCrunch report.