← Archive
Groundswell Post · Sep 6 – 12, 2026

Issue 11

27storiesOrdered by how widely each broke across the AI world — strongest signals first.
01

Model Releases

OpenAI releases ChatGPT Images 2.5

Sourceopenai.com↗

ChatGPT Images 2.5 is an upgraded image-generation model that promises image generation 50% faster than its predecessor.

02

Launches

Meta launches Muse personal AI agent

Sourceauth.muse.ai↗

Meta released Muse, an agent app that connects to email, calendars, shopping, payments, and smart-home devices to perform tasks such as booking tickets and managing an inbox. The app operates through its own background computer, prompting privacy concerns.

03

Industry & Business News

OpenAI reports coding agents now handle 3.1 workdays of research for every human workday

Sourcex.com↗

OpenAI said its coding agents now log 3.1 workdays for every human workday, with about 80% of researchers using at least four agents simultaneously. The company also reported that agent-token spending, output, and experiment volume have risen substantially as tasks become more complex.

04

Research & Papers

OpenAI claims a model produced a 100-page Navier–Stokes proof

Sourceopenai.com↗

OpenAI said its model generated a 100-page proof addressing the Navier–Stokes problem, one of the seven Millennium Prize Problems. The claim has prompted scrutiny over related unpublished drafts allegedly supplied to Codex and questions about their possible use in training.

05

Policy, Safety & Ethics

Anthropic publishes report on misuse of Claude models

Sourceanthropic.com↗

Anthropic’s Threat Intelligence report documents attempted misuse of Claude in areas including surveillance operations and weapons development. It identifies growing autonomy in AI-assisted cyberattacks and a declining need for attackers to have advanced expertise.

22 more stories in this issue

Sign in to read the full issue and every issue in the archive. Free for 14 days.

Open feedback on this issue

Anything about the issue as a whole — a story that should have been here, a category that is off, two entries that are really one event, or how it reads.