Anthropic delays Claude Mythos AI model release over security risks

Anthropic has postponed the public release of its AI model Claude Mythos due to security risks, and odds of Claude 4.7 being released by April 17 now sit at 0.1% YES, down from 32% just 24 hours ago.

Market reaction

The April 17 sub-market collapsed to 0.1% YES. The April 16 market resolved at 99.9% YES, driven by heavy trading on rumors of an imminent release that never materialized. Sub-markets for April 19 and April 21 also sit at 0.1% YES, showing traders don’t expect a near-term release.

Face value trading across these markets tops $215K in 24-hour volume, but actual USDC traded is $132K, a gap between headline figures and real cash at stake. The order book is thin: $258 can shift odds by 5 points, meaning even small trades move the market. The largest single move was a 29-point spike.

Why it matters

Anthropic’s delay stems from specific risks Claude Mythos poses, including its ability to autonomously discover and weaponize software vulnerabilities. This likely extends the release timeline well past the dates traders had been pricing in. A YES share for a May 31 release, currently at 99.9%, offers almost no upside given the risk.

What to watch

Statements from Anthropic CEO Dario Amodei or Co-Founder Jack Clark on the revised timeline. Updates on @anthropicai or official blog posts will be the first signals for reassessing release odds.

Get prediction market intelligence as a structured API feed. Early access waitlist.