You should not install Claude Desktop or Claude Code unless you trust Anthropic. You either trust them to be a responsible custodian of your compute environment or you don't.
I mean it almost doesn't matter what is installed at any given time, the agent is going to install stuff you can't realistically observe, the software will auto-update, there is simply no way you can be sure spyware won't end up on your computer.
That’s not spyware, that’s just how native messaging is designed to work. You have to put a manifest there if you want the native messaging to work later.
Yeah, this. 1password does the same thing for any browser it detects when installed for the native desktop integration from the chrome extension.
Not 100% across the spec but this wouldn't functionally do anything until you install the related extension? e.g., it's pinned to nominated `allowed_origins`
With this argument you could also justify: "That's not a remote access trojan (RAT), that's just how client-server communication is designed to work."
> You have to put a manifest there if you want the native messaging to work later.
The point is that Claude Desktop didn't ask the user whether they want native messaging in the first place. Which is strange, given that users experience many "Do you grant permission to do XYZ" prompts when working with Anthropic products in other situations.
The later parts of this article listing out the dark patterns and security issues and privacy issues is great. Spyware may not be the right term but there is a lot that is wrong here and Anthropic absolutely should be called out for it. Many people and businesses are trusting what appears to be a mix of vibe coded slop and aggressive anti user growth hacks. So much for Anthropic’s high and mighty moralizing.
I don't know why this is flagged, just ran a query on my Mac and indeed, the anthropic extension was deployed for all sorts of (installed and imaginary) chromium browsers.
I've never seen or approved a prompt from Claude if I want any of this to be installed and I've never seen or approved a prompt from macOS that Claude is asking permission to mess around with other apps (though `Application Support` is probably not protected for non-sandboxed apps).
I don't think we should normalise or try to diminish the importance of good security practices. Apps that randomly rewrite how other apps your computer work are generally in the category of malware (and here we're not even considering Claude's apparently ability to execute local instructions based on random text it finds online).
I also checked Claude Desktop > Settings > Extensions. Not a single word or mention of these aforementioned extensions for browsers. I have zero Claude Desktop extensions installed and, without reading the article, would have never guessed that these extensions for browsers were installed.
Claude Desktop repeatedly installed/updated these 7 extensions since the beginning of February on my Apple machine. Every entry in the filtered log below is for all 7 extensions:
I would have thought this would be more of a story, I'm surprised there are so few comments.
"Claude Desktop, an Anthropic application, reached across the trust boundary between two independent vendors, and wrote configuration into Brave's application directory. The principle that an application does not silently modify another application is so obvious it rarely gets stated. Anthropic broke it in silence."
This is the key point for me - ask me, let me remove when done. That would be all it takes to not abuse trust.
17 comments
[ 4.7 ms ] story [ 41.7 ms ] threadI mean it almost doesn't matter what is installed at any given time, the agent is going to install stuff you can't realistically observe, the software will auto-update, there is simply no way you can be sure spyware won't end up on your computer.
"All or nothing" thinking...: https://en.wikipedia.org/wiki/Splitting_%28psychology%29?use...
https://github.com/anthropics/claude-code/issues/14616
Of course if they actually did it, without your consent, that's really really bad.
Not 100% across the spec but this wouldn't functionally do anything until you install the related extension? e.g., it's pinned to nominated `allowed_origins`
> You have to put a manifest there if you want the native messaging to work later.
The point is that Claude Desktop didn't ask the user whether they want native messaging in the first place. Which is strange, given that users experience many "Do you grant permission to do XYZ" prompts when working with Anthropic products in other situations.
That’s not new by any means. It’s no different than zoom installing a native audio plugin for the os, or other software creating files on the system.
I've never seen or approved a prompt from Claude if I want any of this to be installed and I've never seen or approved a prompt from macOS that Claude is asking permission to mess around with other apps (though `Application Support` is probably not protected for non-sandboxed apps).
I don't think we should normalise or try to diminish the importance of good security practices. Apps that randomly rewrite how other apps your computer work are generally in the category of malware (and here we're not even considering Claude's apparently ability to execute local instructions based on random text it finds online).
Claude Desktop repeatedly installed/updated these 7 extensions since the beginning of February on my Apple machine. Every entry in the filtered log below is for all 7 extensions:
"Claude Desktop, an Anthropic application, reached across the trust boundary between two independent vendors, and wrote configuration into Brave's application directory. The principle that an application does not silently modify another application is so obvious it rarely gets stated. Anthropic broke it in silence."
This is the key point for me - ask me, let me remove when done. That would be all it takes to not abuse trust.