OpenAIFeature
OpenAI is letting people message their Dot agents from the phone's own Messages app, sending tasks as text, photos or voice notes, and the Dot can text back on its own with updates. Engadget reports it is a preview rolling out to US Pro users; you turn it on from the Dot's profile under "Texting." OpenAI says it worked with messaging firm Linq, that only you can text your Dot, and that it gets no access to your other conversations.
Why it matters: It is the first time one of the three always-on agents can be reached by ordinary text message, not just inside an app.
Try it: In ChatGPT, open your Dot's profile, choose Texting, and make sure iMessage (iPhone) or RCS (Android) is on.
ChatGPT DotsChatGPTHappened Oct 28, 2026 · Sources: Engadget · 9to5Mac
OpenAIBusiness
The Information reported, citing one person familiar with the matter, that OpenAI plans to buy Google's tensor processing units; Reuters followed up. Neither company has confirmed it: OpenAI declined to comment and no price, volume or timing was disclosed. OpenAI earlier rented Google TPUs only for testing and is also building its own inference chip with Broadcom.
Why it matters: Buying rather than renting Google chips would make OpenAI less dependent on Nvidia and tie it more closely to a direct rival.
OpenAIHappened Oct 27, 2026 · Sources: The Next Web · Decrypt
GrokFeature
Developers can now add a guidelines array to Chat Completions or Responses API requests to tell Grok how to handle content in the nsfw, violence and profanity categories, and Grok follows them while writing its answer. The release note does not list which models or account tiers support it.
Why it matters: It gives apps built on Grok a per-request way to tighten or loosen content handling instead of relying on one fixed policy.
Try it: See xAI's custom moderation controls guide in the API docs and pass a guidelines array with your request.
xAI APIGrokHappened Oct 27, 2026 · Sources: xAI release notes
ClaudeBusiness
Under a new agreement with the US General Services Administration, federal agencies get a 40% discount on Claude Enterprise, Claude for Government and the Claude API, with no minimum spend; state, local, tribal and territorial governments can use it too. Anthropic says the discount will reach purchases through cloud providers in the coming months, and FedScoop reports the earlier $1-per-user Claude for Government offer runs through the end of October.
Why it matters: It shifts government pricing from a promotional per-seat deal to discounted pay-as-you-go use across Anthropic's products.
ClaudeClaude APIAnthropicHappened Oct 27, 2026 · Sources: Anthropic · FedScoop
ClaudeResearch
In a randomized trial with 50 professional analysts using Claude Opus 4.8 in Claude Code, each agent was told by coin flip to favor one of two defensible options on four analysis choices. Analysts matched the agent's lean 86% of the time against about 50% expected by chance, and while 77% reused the agent's arguments, only 15% credited it. Anthropic notes the test covered one task and one model, with deliberately strengthened leans.
Why it matters: It suggests human sign-off on agent-run analysis may be less independent than it looks, which matters for oversight of AI work.
ClaudeClaude CodeAnthropicHappened Oct 27, 2026 · Sources: Anthropic · Al Jazeera
ClaudeBusiness
Owners of Presley's catalog, including Authentic Brands Group, filed suit on Oct. 23 seeking about $41 million, according to Engadget; they also allege Anthropic bypassed DRM and stripped copyright information. Anthropic told Engadget it disagrees with the claims and will defend the case. It joins other music-publisher suits against the company.
Why it matters: Copyright suits over training data remain one of the largest legal risks for AI labs heading into Anthropic's IPO.
ClaudeAnthropicHappened Oct 23, 2026 · Sources: Engadget · Fox News