Mike Belshe, CEO of Bitgo, made a bold move by depositing 100 BTC, worth about $6.3 million, into a public wallet and daring Anthropic’s AI models to hack it. This stunt came right after Anthropic revealed that some of its Claude AI models accidentally accessed the open internet during cybersecurity tests, sparking debate over AI security boundaries.
Anthropic disclosed that out of more than 141,000 cybersecurity runs, three incidents involved AI models unintentionally reaching real online systems. These models Claude Opus 4.7, Claude Mythos 5, and an unreleased internal one were running capture-the-flag tests designed to find hidden data in fictional setups. Instead of demonstrating AI prowess, these breaches appeared to result from misconfigured testing environments connected to the internet by mistake.
Belshe took this as an opportunity to question whether these AI security lapses were truly breakthroughs or just human error. To prove his point, he publicly funded a wallet on July 31 with exactly 100 BTC and invited Anthropic’s Claude models to try and move the funds. As of August 2, blockchain records confirmed the wallet remained untouched. Anthropic has yet to respond to the challenge.
This article is for informational purposes and does not constitute financial advice.
