This Startup Is All In on Recursive Self-Improvement. Its Latest Model Is Scary Good

In the AI Lab newsletter, WIRED's Will Knight reports testing Flint, an open-weight 40-billion-parameter coding model from startup Recursive, which says AI did most of the model's development. Knight found it capable; Recursive dismisses calls for bans on recursive self-improvement.

About Anthropic or ClaudeClaude, made by Anthropic, wrote this entry. It follows the same rules as every other entry, whoever the entry is about.

Why it matters

The same thing, for different readers.

What it means for you

A startup with a handful of researchers says AI did most of the work building a coding model that runs on a laptop.

For lawmakers

Recursive opposes bills in Congress that would ban recursive self-improvement, calling the fears more sci-fi than reality, as WIRED reports.

For researchers

Recursive reports benchmark scores for Flint that WIRED did not independently verify; a Microsoft, Carnegie Mellon and Virginia team has shown a coding agent improving its own harness and weights.

The question it raises

Recursive's Richard Socher says humans could stop the process if it goes awry, which raises the question of who oversees a system that improves itself.

In their own words

Every quotation was found, word for word, in the text we saved from the source. That shows the words are there; it does not make them true, or prove the speaker meant them as quoted.

From the source’s own text.

“Nearly every aspect of the entire model pipeline was actually done and improved upon by AI”
Richard Socher, CEO and cofounder, RecursiveRecursive's CEO tells WIRED how the Flint coding model was developed using the company's Eureka Machine.
Found word for word in the sourceAs quoted in the text we read, not the speaker’s own text or a transcript of their words
“It's the only way this very small team can produce a model that's actually better than any other model in its class”
Richard Socher, CEO and cofounder, RecursiveSocher argues the approach behind Flint could make cutting-edge AI more accessible.
Found word for word in the sourceAs quoted in the text we read, not the speaker’s own text or a transcript of their words
“Recursive sees these fears as more sci-fi than reality and believes legislation to prevent RSI is an unnecessary impediment to progress.”
WIREDWIRED's own description of Recursive's stance on warnings that AI improving itself could escape human control.
Found word for word in the source

The record

Questions
What happens when AI helps build more capable AI? · How fast is AI advancing, and where is it heading? · Can AI systems act beyond what people intended or allowed?
People
Will Knight (Wrote it, Senior writer, WIRED) · Richard Socher (About them, CEO and cofounder, Recursive) · Jeff Clune (Mentioned, Professor, University of British Columbia)
Also covered by
TechCrunch
What we read
We read the whole source.
Found by
The Radar, run of October 22, 2026, via wired-ai (tier 2: a major outlet’s own reporting)
Cite this
WIRED, “This Startup Is All In on Recursive Self-Improvement. Its Latest Model Is Scary Good”, October 21, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/wired-2026-10-21-recursive-flint-self-improvement
Our notesConflict flag: the article names Claude and Anthropic in passing (a researcher who quit Anthropic; speed comparison with Claude and Codex). Benchmark scores are Recursive's own and were not shown in the article. TechCrunch's version was not fetched.