More than 100,000 documents have been exposed online because of a security flaw in Claude Code, the AI-powered coding assistant developed by Anthropic. Among the exposed materials, more than 1,000 instances containing passwords have been identified, along with private company data.
The issue was identified by cybersecurity company Apiiro. According to the company, it stems from the way Claude Code automatically stores users’ preferences and permissions. When files containing this information are copied to or configured on other devices, confidential data can end up unprotected online.
Të lidhura
None found
The information can also appear on GitHub, a platform where programmers and companies store and organize code for their software, applications and technology projects. Experts warn that exposed data may remain accessible online even after the original files are deleted from a device.
Claude Code is designed to read, plan and carry out coding tasks autonomously, without constant human supervision. The case has raised fresh concerns about the security of AI tools and the protection of sensitive data.
Anthropic has said it is aware of the issue and is working on a solution to prevent further exposure of private information.
