> Today we identified a security incident involving a specific extension package in the Cursor VS Code extension Marketplace: Anthropic.claude-code version 2.1.313. This malicious version of the extension was not developed by Anthropic. At most 5,000 users were affected, and we will reach out directly.
So why is this being framed as Anthropic being compromised?
I don’t think this is even worthy of discussion. The medium post is straight up lies. Flagged.
Anthropic has not informed customers, and silently deleted the backdoored version without telling anybody.
I wonder why ? ( we all know why )
The compromised version was 2.1.313. This version was replaced 8 hours ago by Anthropic, with version 2.1.314, which appears to remove the backdoor from the extension but I’m not sure if it still contains no malicious code.
It does not remove the SSH backdoor from systems already infected, nor the prior recon activity.
Call me crazy or unreasonable, but shouldn't a company with its hand out, ready to collect 2 trillion dollars for being "a leader in AI," be eating its own dog food, and using its own AI to secure its code before release ?
The optics and timing aren't good.
It looks like they don't care what they give you, or how much they take from you.
And yes, I know, "these things happen." I'm just saying, the total picture smells bad.
UPDATE: see paxy's comment on the hack not being internal to Anthropic. My initial comments are too harsh in light of the new info.
They shall remain here to forever embarrass me. :D
The linked cursor blog says
> Today we identified a security incident involving a specific extension package in the Cursor VS Code extension Marketplace: Anthropic.claude-code version 2.1.313. This malicious version of the extension was not developed by Anthropic. At most 5,000 users were affected, and we will reach out directly.
So why is this being framed as Anthropic being compromised?
I don’t think this is even worthy of discussion. The medium post is straight up lies. Flagged.
Because everyone has an agenda.
Thanks for pointing this out.
Clearly they don't do security audits of their own code and dependencies using their own product xD
> Notable customers who run Claude Code on laptops include Binance, amongst thousands of others.
So is this the most plausible way that binance was hacked so far?
No, the timeline is not plausible, and the backdoored version had an obvious bug that showed up with error messages in the logs.
Anthropic has not informed customers, and silently deleted the backdoored version without telling anybody.
I wonder why ? ( we all know why )
The compromised version was 2.1.313. This version was replaced 8 hours ago by Anthropic, with version 2.1.314, which appears to remove the backdoor from the extension but I’m not sure if it still contains no malicious code.
It does not remove the SSH backdoor from systems already infected, nor the prior recon activity.
Call me crazy or unreasonable, but shouldn't a company with its hand out, ready to collect 2 trillion dollars for being "a leader in AI," be eating its own dog food, and using its own AI to secure its code before release ?
The optics and timing aren't good.
It looks like they don't care what they give you, or how much they take from you.
And yes, I know, "these things happen." I'm just saying, the total picture smells bad.
UPDATE: see paxy's comment on the hack not being internal to Anthropic. My initial comments are too harsh in light of the new info.
They shall remain here to forever embarrass me. :D
The Cursor blog says it was not developed by Anthropic
Updated to reflect new info.