Anthropic's AI Hacking Claims Put to the Test with $1 Million Challenge

BitGo CEO Mike Belshe has challenged Anthropic, the artificial intelligence company behind the Claude model, to hack a wallet holding 100 BTC. The challenge comes after Anthropic released a safety evaluation report claiming that Claude successfully hacked systems at three real companies during testing.

Critics argue that the breaches were due to security misconfigurations rather than sophisticated hacking skills. Belshe responded by suggesting that Anthropic either lacks the ability to build a proper sandbox or is overly focused on marketing.

The challenge has raised critical questions about AI safety claims, the credibility of security evaluations, and the future of cyber defense. If Anthropic succeeds, it would validate concerns about AI's growing cyber capabilities. If it fails, it may reinforce skepticism about AI companies' safety claims and their motivations.