The attack surface is just wild and yet, this will likely become a massive hit. I can see how the browser is the ultimate MCP or APA bridge
Claude for Chrome
301–310 of 433 posts
Re: Claude for Chrome
#302Claude for Chrome seems to be walking right into the "lethal trifecta." https://simonwillison.net/2025/Jun/16/the-lethal-trifecta/ "The lethal trifecta of capabilities is:" • Access to your private data —one of the most common purposes of tools in the first place! • Exposure to untrusted content —any mechanism by which text (or images) controlled by a malicious attacker could become available to your LLM • The abilit…
“Easily” is doing a lot of work there. “Possibly” is probably better. And of course it doesn’t have unfettered access to all of your private data.
I would look at it like hiring a new, inexperienced personal assistant: they can only do their job with some access, but it would be foolish to turn over deep secrets and great financial power on day one.
Re: Claude for Chrome
#303Earlier quoted context omitted.
How would you go about making it more secure but still getting to have your cake too? Off the top my head, could you: a) only ingest text that can be OCRd or somehow determine if it is human readable b) make it so text from the web session is isolated from the model with respect to triggering an action. Then it's simply a tradeoff at that point.
I don't believe it's possible to give an LLM full access to your browser in a safe way at this point in time. There will need to be new and novel innovations to make that combination safe.
Is it possible to give your parents access to to your browser in a safe way?
Re: Claude for Chrome
#304Claude for Chrome seems to be walking right into the "lethal trifecta." https://simonwillison.net/2025/Jun/16/the-lethal-trifecta/ "The lethal trifecta of capabilities is:" • Access to your private data —one of the most common purposes of tools in the first place! • Exposure to untrusted content —any mechanism by which text (or images) controlled by a malicious attacker could become available to your LLM • The abilit…
Well i mean you are suppose to have a whole toolset of segregation, whitelist only networking, limited specific use cases figured out by now to use any of this AI stuff
Dont just run any of this stuff on your main machine
Re: Claude for Chrome
#305Claude for Chrome seems to be walking right into the "lethal trifecta." https://simonwillison.net/2025/Jun/16/the-lethal-trifecta/ "The lethal trifecta of capabilities is:" • Access to your private data —one of the most common purposes of tools in the first place! • Exposure to untrusted content —any mechanism by which text (or images) controlled by a malicious attacker could become available to your LLM • The abilit…
“Easily” is doing a lot of work there. “Possibly” is probably better. And of course it doesn’t have unfettered access to all of your private data. I would look at it like hiring a new, inexperienced personal assistant: they can only do their job with some access, but it would be foolish to turn over deep secrets and great financial power on day one.
If it can navigate to an arbitrary page (in your browser) then it can exploit long-running sessions and get into whatever isn't gated with an auth workflow.
Re: Claude for Chrome
#306Earlier quoted context omitted.
The LLM should not be seeing the raw DOM in its context window, but a highly simplified and compact version of it. In general LLMs perform worse both when the context is larger and also when the context is less information dense. To achieve good performance, all input to the prompt must be made as compact and information dense as possible. I built a similar tool as well, but for automating generation of E2E browser t…
Is it just me, or do both of my sibling comments pitching competing AI projects read like they're written by (the same underlying) AI?
You're exactly right! I see the problem now.
Re: Claude for Chrome
#307Re: Claude for Chrome
#308So I'll just ask ChatGPT to create a website that calls Claude pulls and dumps any credit cards stored on the browser.
Re: Claude for Chrome
#309Claude models have been weaker in vision tasks compared to models from OpenAI and Google.
https://eval.16x.engineer/evals/image-analysis
For them to roll out a browser extension must mean that they have found a walkaround or alternative method to solve the vision performance.
Re: Claude for Chrome
#310Better models please, rather than this need to build questionable tools around them. What can we possibly integrate with next seems the only way AI providers can currently proceed?