Project Glasswing Archives

Inside the race to adapt to an AI-powered security world

AI is breaking things faster than anyone can fix them. Security leaders across the industry are racing to figure out what comes next.

3 days ago By Greg Otto

Governments and financial institutions are reviewing potential cybersecurity risks from the company’s advanced AI model, which has identified thousands of software vulnerabilities but is restricted over misuse concerns. (Photo Illustration by Jonathan Raa/NurPhoto via Getty Images)

Anthropic expanding access to Project Glasswing

Roughly 150 new organizations across critical infrastructure sectors will gain access to Claude Mythos Preview, Anthropic's most capable — and most restricted — AI model.

5 days ago By Greg Otto

Several partners reported that their rates of bug discovery with Mythos had increased more than tenfold. (Getty Images)

Anthropic: Mythos finds more than 10,000 software flaws in first month

Early results show a tenfold jump in bug discovery at some partners, and a widening gap between finding flaws and fixing them.

May 26, 2026 By Greg Otto

Researchers say AI just broke every benchmark for autonomous cyber capability

Two independent studies found that Anthropic's Claude Mythos Preview and OpenAI's GPT-5.5 have outpaced every trend line researchers were tracking. No one is sure if this is…

May 13, 2026 By Greg Otto

Anthropic’s Project Glasswing website is displayed on a smartphone screen in this photo illustration in Brussels, Belgium, on April 12, 2026. Governments and financial institutions are reviewing potential cybersecurity risks from the company’s advanced AI model, which has identified thousands of software vulnerabilities but is restricted over misuse concerns. (Photo Illustration by Jonathan Raa/NurPhoto via Getty Images)

Here’s how cyber heavyweights in the US and UK are dealing with Claude Mythos

Two reports from former high-level U.S. cyber officials and the UK government’s top AI research institution reveal how top defenders think about the tool’s hacking capabilities.

Apr 13, 2026 By Derek B. Johnson

Project Glasswing, named after the Greta oto butterfly, aims to secure open-source software with the use of a new frontier model. (Getty Images)

Tech giants launch AI-powered ‘Project Glasswing’ to identify critical software vulnerabilities

The program comes as the tech industry races to secure software before similar AI-powered offensive capabilities become too much for defenders to handle.

Apr 7, 2026 By Greg Otto