Anthropic's Claude Code Tool Hits Critical Vulnerability, Users at Risk of System Crashes

Anthropic's code assistant tool may have been highly anticipated, but the current crisis has left some users scratching their heads. According to GitHub users, there are some serious flaws in the auto-update function. Under specific conditions, this vulnerability could render users' workstations unstable or completely inoperable, effectively "bricking" them.

Early investigations reveal that the problem primarily affects systems where Claude Code is installed with "root" or "superuser" privileges. These elevated permissions allow the program to modify low-level operating system settings. When the flawed auto-update command executes, the application can modify directories typically under strict protection. This could prove harmful to operating systems. In the worst-case scenario, these erroneous commands could completely corrupt the system, preventing it from booting—the so-called "bricking" effect.

One GitHub user claimed to have to boot a "rescue instance" to fix file permissions inadvertently damaged by a Claude Code command. The core issue lies in the auto-update command incorrectly altering the access permissions of crucial system files. Since file permissions determine which programs and users can read, modify, or execute specific applications, disrupted permissions will lead to system malfunctions.

What has Anthropic Done? 

To their credit, Anthropic has responded swiftly to user feedback. The AI leader took immediate action by removing the problematic auto-update command from Claude Code. Also, they added a link within the program directing users to a troubleshooting guide. Notably, this link initially contained a spelling error, which Anthropic claims to have corrected.

There's no formal or official statement from Anthropic regarding the extent of the impact of any potential user data security concerns.