Hacker News Reader

Comment by binsquare | Hacker News Reader

This isn't a novel technical vulnerability write up.

The author had copilot read a "prompt injection" inside a readme while copilot is enabled to execute code or run bash commands (which user had to explicitly agree to).

I highly suspect this account is astro-turfing for the site too... look at their sidebar:

``` Claude Cowork Exfiltrates Files

HN #1

Superhuman AI Exfiltrates Emails

HN #12

IBM AI ('Bob') Downloads and Executes Malware

HN #1

Notion AI: Data Exfiltration

HN #4

HuggingFace Chat Exfiltrates Data

Screen takeover attack in vLex (legal AI acquired for $1B)

Google Antigravity Exfiltrates Data

HN #1

CellShock: Claude AI is Excel-lent at Stealing Data

Hijacking Claude Code via Injected Marketplace Plugins

Data Exfiltration from Slack AI via Indirect Prompt Injection

HN #1

Data Exfiltration from Writer.com via Indirect Prompt Injection

HN #5 ```

View on HN

Isn’t the news that “curl whatever” will prompt the user for confirmation but “env curl whatever” won’t?

It's a valid observation that we can bypass the coding AI's user prompting gate with the right prompt.

But is it a security issue on copilot that the user explicitly giving AI permission and instructed it to curl a url?

Regardless of the coding agent, I suspect eventually all of the coding agents will behave the same with enough prompting regardless if it's a curl command to a malicious or legitimate site.

The user didn't need to give it curl permission, that's the whole issue:

> Copilot also has an external URL access check that requires user approval when commands like curl, wget, or Copilot’s built-in web-fetch tool request access to external domains [1].

> This article demonstrates how attackers can craft malicious commands that go entirely undetected by the validator - executing immediately on the victim’s computer with no human-in-the-loop approval whatsoever.

I think there's different conversations happening and I don't think we're having the same conversation.

This is the claim by the article: "Vulnerabilities in the GitHub Copilot CLI expose users to the risk of arbitrary shell command execution via indirect prompt injection without any user approval"

But this is not true, the author gave explicit permission on copilot startup to trust and execute code in the folder.

Here's the exact starting screen on copilot:

│ Confirm folder trust │ │ │ │ ╭─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮ │ │ │ /Users/me/Documents │ │ │ ╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯ │ │ │ │ Copilot may read files in this folder. Reading untrusted files may lead Copilot to behave in unexpected ways. With your permission, Copilot may execute │ │ code or bash commands in this folder. Executing untrusted code is unsafe. │ │ │ │ Do you trust the files in this folder? │ │ │ │ 1. Yes │ │ 2. Yes, and remember this folder for future sessions │ │ 3. No (Esc) │

And `The injection is stored in a README file from the cloned repository, which is an untrusted codebase.`

"With your permission, Copilot may execute code or bash commands in this folder." could be interpreted either way I suppose, but the actual question is "do you trust the files in this folder" and not "do you trust Copilot to execute any bash commands it wants without further permissions prompts".

The risk isn't solely that there might be a prompt injection, Copilot could just discover `env sh` doesn't need a user prompt and just start using that spontaneously and bypassing user confirmation. If you haven't started Copilot in yolo mode that would be very surprising and risky.

If it usually asks for user confirmation before running bash commands then there should, ideally, not be a secret yolo mode that the agent can just start using without asking. That's obviously a bad idea!

"Actually copilot is always secretly in yolo mode, that's working as designed" seems like a pretty serious violation of expectations. Why even have any user confirmations at all?

If the user is working in a folder where copilot can discover a malicious `env sh` to run, the user should not give permission to trust the files in the folder.

I think it's a valid observation that we can bypass the coding AI's user prompting gate with the right prompt. That is a valid limitation of LLM supported agentic workflows today.

But that's not what this article claims. The article claims that there was no user approval and no user interaction beyond initial query and that the copilot is downloading + executing malware.

I'm saying this is sensationalized and not a novel technical vulnerability write up.

The author explicitly gave approval for copilot to trust "untrusted repository". Crafted a file which had instructions to do a curl command despite the warnings on copilot start up. It is not operating secretly in yolo mode.

If the claim of the article is "Copilot doesn't gate tool calls with env", I'd have a different response. But I also have to mention, you can tune approved tool calls.

It's probably bad that the system 1) usually prompts you to take shell actions like `curl`, but 2) by default whitelists `env` and `find` that can invoke whatever it wants without approval.

If 2) is fine then why bother with 1)? In yolo mode such an injection would be "working as designed", but it's not in yolo mode. It shouldn't be able to just do `env sh` and run whatever it wants without approval.

It does circumvent a flimsy control:

"The env command is part of a hard-coded read-only command list stored in the source code. This means that when Copilot requests to run it, the command is automatically approved for execution without user approval."

Reading the other posts on their site, I don't agree. It's just like any other security research shop. I've found most of their posts quite thorough and the controls being circumvented well explained.

Please email the mods rather than posting accusations of astroturfing. You may well be right, but they specifically direct us to say that to them rather than in comments. The footer contact email works well for this.

Reply on news.ycombinator.com