In Cybersecurity, Claude Leaves Other LLMs in the Dust

Anthropic proves that LLMs can be fairly resistant to abuse. Most developers are either incapable of building safer tools, or unwilling to invest in doing so.