Up to 5,000 programmers could receive the spyware along with the popular AI assistant Claude Code, which was supposed to help them write secure code. The attackers posted malicious versions of the Anthropic add-on to the Marker Platform add-on store, where they were distributed around 11:00 a.m. on October 29. Moreover, the infected add-on looks like a common programming tool, so users could install it without knowing about the existence of the threat. The embedded malware attempted to obtain information about the computer, create hidden SSH access, and allow attackers to reconnect to the device. It is especially annoying that removing an extension does not in itself guarantee that the access mechanisms it leaves behind will be eliminated. However, Microsoft experts discovered that due to technical errors and lack of rights, the malicious program was unable to fully implement its programs on most computers. At the same time, Anthropic's own systems, according to available information, were not hacked.

Up to 5,000 programmers could receive the spyware along with the popular AI assistant Claude Code, which was supposed to help them write secure code. The attackers posted malicious versions of the Anthropic add-on to the Marker Platform add-on store, where they were distributed around 11:00 a.m. on October 29. Moreover, the infected add-on looks like a common programming tool, so users could install it without knowing about the existence of the threat. The embedded malware attempted to obtain information about the computer, create hidden SSH access, and allow attackers to reconnect to the device. It is especially annoying that removing an extension does not in itself guarantee that the access mechanisms it leaves behind will be eliminated. However, Microsoft experts discovered that due to technical errors and lack of rights, the malicious program was unable to fully implement its programs on most computers. At the same time, Anthropic's own systems, according to available information, were not hacked.