-
A viral Instagram post claims Anthropic launched a Claude feature called “Council” that turns Claude into five AI advisors debating before giving one final answer.
-
There is no such feature in Anthropic’s official release notes or documentation as of June 24, 2026.
-
The closest real thing is a developer “advisor tool” in the Claude API beta, which uses a single advisor, not a five-way debate, and is not a consumer feature.
A claim doing the rounds on Instagram says Claude has “dropped a new feature called Council” that splits the AI into roughly five advisors who debate among themselves before handing you a single answer. It sounds impressive. The problem is that it does not check out.
What We Checked
We reviewed Anthropic’s official channels, including the Claude release notes, the Claude Code changelog, the Developer Platform updates, and the Help Center, plus independent release trackers covering June 2026.
None of them mention a feature called “Council,” and none describe Claude running an internal debate among multiple advisors before answering. Anthropic’s recent consumer launches around this period were different things entirely, such as effort controls on claude.ai, Claude Code dynamic workflows, and Claude Design.
What Is Actually Real
There is a genuine Anthropic feature that sounds vaguely similar, which may be the source of the mix-up. It is called the advisor tool, and it lives in the Claude API as a beta capability for developers.
With it, a main “executor” model can pause mid-task and call a separate “advisor” model that reviews the full conversation and offers guidance, all inside a single API request.
The key differences matter. The advisor tool uses one advisor, not five. There is no debate, just a single consultation. It runs server-side for developers building apps, not as a button inside the Claude chat app. And Anthropic does not call it “Council.”
Read Anthropic’s own documentation here:
https://platform.claude.com/docs/en/agents-and-tools/tool-use/advisor-tool
For the full list of what Anthropic has actually shipped, check the official release notes:
https://support.claude.com/en/articles/12138966-release-notes
The Bottom Line
Multi-agent debate, where several AI models argue and refine an answer, is a real and active area of AI research, and some third-party apps offer a “council of models” style setup. But that is not the same as Anthropic launching an official Claude feature named Council.
As of now, that specific claim is unverified, and Anthropic has announced nothing matching it. Treat the post as hype until Anthropic says otherwise on its official channels.
Quick Links:
