A major revelation has emerged regarding the security systems of leading artificial intelligence company Anthropic. US AI startup Cathedral has claimed that it gained access to some of the company’s internal systems in less than 12 hours.
According to the US-based Cathedral, the operation was carried out under a bug bounty programme, which aims to identify security vulnerabilities in companies’ systems. Through this programme, external experts are given the opportunity to find potential weaknesses.
According to a report by the American newspaper The Washington Post, access was gained during the operation to internal messages exchanged on Slack, written records of meetings, employee salaries and information related to job offers.
Anthropic also confirmed the incident, stating that Cathedral had gained access to limited portions of the company’s sales, human resources and information technology systems.
However, the company said that access to users’ conversations with Claude was not obtained and that the source code of its AI model remained secure.
According to Anthropic, the company’s monitoring systems had detected some of Cathedral’s activities even before the incident was reported. The company further stated that the relevant access was disabled on the same day.
The incident highlights the importance of protecting internal systems, employee information and sensitive data for companies developing modern artificial intelligence technologies.
It is worth noting that the operation involved testing security vulnerabilities under a bug bounty programme. Therefore, it is important to understand the full context of the incident before equating it with an unauthorised cyberattack.
