Anthropically warns of new “vibe hacking” attacks used by Claude AI

Antropic, the company behind the popular and Model Clauda, ​​said in a new one Threat report to disrupt the “vibe to hack” the scheme of extortion. The report described in detail how the attack was executed, allowing hackers to increase the mass attack on 17 targets, including entities in government, health care, emergency services and religious organizations.

(You can read the full report In this PDF file.)

Anthropic says his Claude and technology was also used as a “technical consultant and active operator, allowing attacks that would be difficult and longer for individual actors to manually execute manually.” Claude was used to “automate reconnaissance, harvesting for believing and network penetration on a scale,” the report said.

The method of finding disturbing that the so-called Vibe hacking was considered a future threat, and some experts believe it was It’s not yet possible. What is anthropic shared in your report can they represent a big shift In how models and agents are used to scale massive cicematists, redemption or fraud schemes.

Separately anthropic recently dealt with other AI problems, Namely, the authors’ lawsuit are solved Claud’s claim is trained on their copyrighted materials. Another company, confusion, dealt with one’s own security issues such as his case AI browser shown to have a great vulnerability.