Can Anthropic Keep Its Exploit-Writing AI Out of the Wrong Hands?

Anthropic has unveiled its new Mythos Preview model, which possesses the concerning ability to discover and exploit critical zero-day vulnerabilities in software. The company insists the powerful AI comes with built-in controls designed to prevent misuse by bad actors. This development raises significant questions about whether these safeguards will be sufficient to keep such exploit-writing capabilities from falling into the wrong hands, as cybersecurity experts debate the risks of releasing tools that could automate the discovery of previously unknown security flaws.