Cathedral, a months-old cybersecurity startup founded by ex-DOGE staffers, says it breached Anthropic's internal systems in under 12 hours using AI agents

Your AI assistant's access to your internal systems may now be a security vulnerability.

Cathedral, a cybersecurity startup founded by former members of the hacking group DOGE, claims to have breached the internal systems of AI company Anthropic within 12 hours. The startup, which is only a few months old, reported its findings to Anthropic, the creator of the Claude chatbot. Cathedral stated that its researchers utilized AI agents to achieve this rapid penetration of sensitive systems. The group's background with DOGE, known for its offensive cybersecurity operations, adds a layer of concern to this reported incident. The specifics of the breach and the exact nature of the systems accessed by Cathedral remain under review.

This reported breach by Cathedral, a new entrant in the cybersecurity space, highlights the escalating capabilities of AI-driven attack vectors against even sophisticated AI companies. The speed at which Cathedral claims to have bypassed Anthropic's defenses suggests that current security measures may not be adequately prepared for autonomous AI agents designed for offensive operations. This incident could accelerate the arms race between AI development and AI security, pushing companies to invest more heavily in defenses specifically tailored to counter AI-powered threats.

The implications extend beyond AI firms, potentially affecting any organization that integrates AI agents into its workflows or grants them access to internal data. If AI agents can be weaponized to exploit the very systems they are designed to assist, the trust and security paradigms surrounding AI adoption will need a fundamental reassessment. Businesses will likely face increased pressure to implement stringent AI governance and security protocols to mitigate the risks posed by these emerging adversarial capabilities.