Skip to content
IIAIT

AI Today · 5 November 2026

Meta buys the Conductor coding-agent app, and OpenAI reports 948 AI-built Linux exploits

A coding tool used by many developers is closing its cloud service within weeks, an OpenAI security run shows how quickly AI can now turn kernel bugs into working attacks, ChatGPT gains animated GIFs, and Anthropic describes a lab project run largely by Claude.

  1. 01

    Meta acquires Conductor; its cloud service for coding agents shuts down on 20 November

    Conductor, an app that lets developers run several AI coding agents such as Claude Code and Codex in parallel workspaces, announced on 4 November 2026 that it is joining Meta. Meta confirmed the deal to TechCrunch but did not disclose terms. According to Conductor, Conductor Cloud becomes read-only on 13 November and shuts down on 20 November, after which TechCrunch reports the company will begin deleting cloud data. The local Mac app will keep receiving new models and critical fixes until early 2027.

    What it means for you

    If your team uses Conductor Cloud, export your data before 13 November: Conductor points to Settings, Account, Export data, or Transfer to local for individual workspaces. It is also a reminder to check, for every AI tool you depend on, how you would get your work out if the service closed.

  2. 02

    OpenAI says an automated AI pipeline produced 948 working Linux root exploits in about 30 hours

    OpenAI has described a mid-October 2026 run in which an upcoming, not yet released cyber-focused model analysed 2,308 candidate Linux kernel findings, reproduced 1,724 vulnerabilities and produced 948 working local privilege escalation exploits, plus 291 information-leak proofs of concept, The Hacker News reports. The run cost 1,213,430 dollars in API credits, about 1,279 dollars per exploit. The pipeline also drafts patches, and security researchers review each finding before it goes to the kernel maintainers.

    What it means for you

    Finding and weaponising known bugs is becoming cheap, so the time between a fix being published and attacks using it is likely to shrink. Ask whoever runs your Linux servers and cloud machines how quickly kernel updates are applied, and limit who has even low-level accounts on them.

  3. 03

    ChatGPT can now create and edit animated GIFs with OpenAI's Images 2.6 model

    OpenAI released Images 2.6 on 4 November 2026. In ChatGPT and Codex on web, desktop and mobile, users can turn a text prompt or an uploaded photo into an animated GIF, change its playback speed and edit individual frames. OpenAI says the model also makes image generation faster. For developers there are two API variants: GPT-Image-2.6 Flare, aimed at speed, and GPT-Image-2.6 Sunburst, aimed at quality. The rollout is gradual, so not every account may see the feature yet.

    What it means for you

    Short animations for presentations, product demos or social posts can now be drafted in ChatGPT without a separate tool. Check the result before publishing, and do not upload photos of colleagues or customers without their consent.

  4. 04

    Anthropic reports a protein structure found with Claude doing much of the lab and analysis work

    Anthropic's biology research lab published on 4 November 2026 the first 3D structure of a vault particle from prokaryotes, a large protein shell whose function is still unknown. Under the scientists' direction, Claude searched a database of about 15 billion predicted proteins, designed DNA constructs and ran the cryo-EM data processing; seven of 15 tested proteins formed vault-like particles. Anthropic also lists Claude's mistakes, including a protocol error that cost two weeks, and says the findings are a preprint submitted to bioRxiv, not yet peer reviewed.

    What it means for you

    AI models can now carry out long chains of technical work, but this example also shows why each step still needs expert review. If you hand multi-step tasks to an AI assistant, set checkpoints where a person confirms the method before the next stage starts.