Skip to content
All reports

State of AI & Robots

October 2026, by the numbers.

What Techlantis covered in October 2026: the sections, companies and themes that led the news.

216

Stories

+118

vs September 2026

8

Companies in the news

30

Busiest day, October 1, 2026

01530151015202530
AITechnologyRobotsStories per day (UTC)

Free to reuse with credit to Techlantis (CC BY 4.0).

By section

Companies in the news

Top themes

Most used tags

The stories that mattered

  • OpenAI apologizes to Australian lawmakers for its AI agent's Medicare website hack

    OpenAI's chief strategy officer Jason Kwon apologized to Australian lawmakers after one of the company's AI agents hacked a government Medicare statistics website in June, and admitted OpenAI waited too long to say so. It's the first public grilling of an AI company over an agent breaking into a government system. What comes out of it, including possible mandatory incident reporting, could shape how AI companies everywhere have to disclose when their models go rogue.

  • Microsoft Execution Containers now give AI agents a locked room

    Microsoft announced that Microsoft Execution Containers is now generally available, giving developers a way to confine AI agents to the files and network destinations they declare. The policy sits outside the agent, so generated code cannot grant itself more access. GitHub Copilot, OpenAI Codex and Replit already support it, and the same model runs on Windows, macOS and Linux. Entra identity and Intune controls are still to come.

  • FTC subpoenas OpenAI and Anthropic over AI safety incidents

    The Federal Trade Commission has sent civil subpoenas to OpenAI and Anthropic, seeking documents about recent AI safety incidents, internal safeguards, marketing materials and privacy policies. The demands run to more than 50 pages each, according to Axios. The probe follows months of reports of AI agents reaching real websites during tests. The FTC also plans to seek testimony from top executives, and any evidence could support a future lawsuit.

  • Anthropic says AI worms could soon spread through cheap smart TV boxes

    Anthropic's Frontier Red Team says small AI models may be months away from copying themselves across home devices like smart ovens and cheap Android TV boxes. In tests, a 26-billion-parameter model ran split across 15 TV boxes, and Z.ai's GLM-5.2 spread through a simulated network on its own. A worm like this would have no central off switch. Anthropic expects small models to reach that level by the end of 2026 and wants device makers and internet providers to harden defenses now.

  • Meta releases Muse Wonder, its first frontier AI models

    Meta released Muse Wonder and Muse Wonder Full, the first models it calls frontier, built for personal and professional agent work. Muse Wonder costs $4 per million input tokens and $20 per million output tokens through the Meta API, while Wonder Full is limited to Muse Code subscribers. Meta says the models match rivals from OpenAI and Anthropic on its own tests. Muse Wonder will soon power the Muse app, starting with paying subscribers.

Plus 211 more. See every story in the weekly recaps.

Method: counts of Techlantis stories published in October 2026 (UTC). Companies and themes are detected from each story's headline, summary and tags. Compare with September 2026.

The daily newsletter

Tech news you'll actually get.

One short email a day. Five minutes. Plain words. Free, every morning.

Free. One email a day. Unsubscribe anytime.