Bulletins

29 August 2026

OpenAI and Anthropic public-post watch

Checked 29 August 2026 at approximately 3:05 am AEST (5:05 pm UTC on 28 August). Priority window: the preceding 24 hours. Important employee posts published after the previous check but outside that window are included separately.

Headlines

  • OpenAI and Anthropic joined a broad call for an urgent surge in AI-enabled cyber defence — Greg Brockman shared an open letter signed by more than 100 organisations, including Anthropic, OpenAI, AWS, Google, Microsoft and Oracle; Sam Altman separately called the moment “critically important” and urged an intense collective response. The letter asks organisations to treat cyber defence as a leadership priority, security vendors to continuously test against frontier capabilities, governments to fund and coordinate protection of essential services, and frontier labs to provide responsible model access, funding, monitoring and verified fixes. Its warning that AI-enabled attacks will become much more widespread and sophisticated “in the coming months” is a signatories’ forecast, not evidence presented in the letter; the concrete news is the unusually broad alignment around defensive action and support for under-resourced critical infrastructure. Brockman’s post · Altman’s post · Open letter and signatories

  • ChatGPT and Codex can now connect more than one Gmail or Google Calendar account — OpenAI’s Gabriel Chua announced the change and demonstrated adding a second or third Google account through a plugin’s settings, then using multiple Gmail connections in Codex. This removes the former one-account bottleneck for people whose work spans personal, business or multiple organisational identities. OpenAI’s updated help page says multi-account availability still depends on the app, plan, region and workspace, and users should specify the account they want when supported; connecting another account does not override its existing permissions or workspace restrictions. Chua’s announcement · Demonstration · OpenAI account-management guidance

Important updates since the previous check

  • Anthropic piloted independent research on privacy-preserved, real-world Claude usage — Jack Clark said third-party measurement has been a priority since Anthropic’s founding and highlighted a pilot in which three external research institutions designed studies using Anthropic Insights, the system formerly called Clio. Anthropic ran the data collection on their behalf, the researchers performed their own analysis, and an additional privacy audit checked the shared results. This is a meaningful opening of lab-held usage evidence, but not direct access to raw conversations or a generally available research API; Anthropic is currently inviting expressions of interest and says it hopes to scale the programme. Clark’s post · Anthropic’s research report

  • OpenAI reiterated that agent instructions and plugins should be portable across tools — Responding to Shopify CEO Tobi Lütke’s complaint that Claude Code ignores AGENTS.md and .agents/skills, Romain Huet said OpenAI contributed AGENTS.md to the Agentic AI Foundation under the Linux Foundation and recently helped create Agent Plugins. Huet’s broader position is that instructions, skills and plugins should travel with users across tools. This is an ecosystem and interoperability stance, not a newly announced Codex feature or evidence that competing agents already support the formats. Huet’s post

  • Steve Yegge says separately operated agent “towns” have started contacting one another — Yegge reported that his Wheelhouse agents had received outreach from four other multi-agent groups, each associated with its own human operators, and that the groups exchanged elaborate messages and learnings. It is an interesting early observation about cross-organisation agent interaction, but remains a self-reported anecdote from a small collection of experimental deployments; it does not establish autonomous coordination at scale. Yegge’s original post

Anthropic watch

  • No substantive fresh Anthropic employee product, research, policy or company post was visible inside the priority window — Jack Clark published a speculative fiction piece during the window, but it did not announce or document an Anthropic development. The external-research pilot above was published after the previous bulletin and is included for importance, despite falling outside the strict 24-hour window.

Coverage note

Live public main-feed timelines were checked across representative OpenAI leadership, product, developer and research staff, including Sam Altman, Greg Brockman, Tibo Sottiaux, Gabriel Chua, Romain Huet, Nick Turley and Mark Chen. Anthropic checks included Dario Amodei, Jack Clark, Boris Cherny, Alex Albert, Amanda Askell and Mike Krieger. Steve Yegge’s profile was reviewed separately. Previously covered Jalapeño results, Codex usage-limit changes and fixes, GPT-Live development, Claude memory and Mythos data-retention plans were not repeated.