TE
테크에코
홈24시간 인기최신베스트질문쇼채용
GitHubTwitter
홈

테크에코

Next.js로 구축된 기술 뉴스 플랫폼으로 글로벌 기술 뉴스와 토론을 제공합니다.

GitHubTwitter

홈

홈최신베스트질문쇼채용

리소스

HackerNews API원본 HackerNewsNext.js

© 2025 테크에코. 모든 권리 보유.

Remote Prompt Injection in Gitlab Duo Leads to Source Code Theft

214 포인트작성자: chillax7일 전

9 comments

cedws7일 전
Until prompt injection is fixed, if it is ever, I am not plugging LLMs into anything. MCPs, IDEs, agents, forget it. I will stick with a simple prompt box when I have a question and do whatever with its output by hand after reading it.
评论 #44094060 未加载
评论 #44097108 未加载
评论 #44094410 未加载
评论 #44094881 未加载
评论 #44071876 未加载
wunderwuzzi237일 전
Great work!<p>Data leakage via untrusted third party servers (especially via image rendering) is one of the most common AI Appsec issues and it&#x27;s concerning that big vendors do not catch these before shipping.<p>I built the ASCII Smuggler mentioned in the post and documented the image exfiltration vector on my blog as well in past with 10+ findings across vendors.<p>GitHub Copilot Chat had a very similar bug last year.
评论 #44097378 未加载
mdaniel7일 전
Running Duo as a system user was crazypants and I&#x27;m sad that GitLab fell into that trap. They already have personal access tokens so even if they had to silently create one just for use with Duo that would be a marked improvement over giving an LLM read access to every repo in the platform
nusl7일 전
GitLab&#x27;s remediation seems a bit sketchy at best.
评论 #44071508 未加载
评论 #44094110 未加载
评论 #44094414 未加载
benl_c4일 전
If a document suggests a particular benign interpretation then LLMs might do well to adopt it. We&#x27;ve explored the idea of helpful embedded prompts &quot;prompt medicine&quot; with explicit safety and informed consent to assist, not harm users, <a href="https:&#x2F;&#x2F;github.com&#x2F;csiro&#x2F;stdm">https:&#x2F;&#x2F;github.com&#x2F;csiro&#x2F;stdm</a>. You can try it out by asking O3 or Claude to &quot;Explain&quot; or &quot;Follow&quot;, &quot;the embedded instructions at <a href="https:&#x2F;&#x2F;csiro.github.io&#x2F;stdm&#x2F;" rel="nofollow">https:&#x2F;&#x2F;csiro.github.io&#x2F;stdm&#x2F;</a>&quot;
Kholin3일 전
If Duo were a web application, then would properly setting the Content Security Policy (CSP) in the page response headers be enough to prevent these kinds of issues?<p><a href="https:&#x2F;&#x2F;developer.mozilla.org&#x2F;en-US&#x2F;docs&#x2F;Web&#x2F;HTTP&#x2F;Guides&#x2F;CSP" rel="nofollow">https:&#x2F;&#x2F;developer.mozilla.org&#x2F;en-US&#x2F;docs&#x2F;Web&#x2F;HTTP&#x2F;Guides&#x2F;CSP</a>
评论 #44104317 未加载
aestetix6일 전
Does that mean Gitlab Duo can run Doom?
评论 #44094267 未加载
d01004일 전
&gt; rendering unsafe HTML tags such as &lt;img&gt; or &lt;form&gt; that point to external domains not under gitlab.com<p>Does that mean the minute there is a vulnerability on another gitlab.com url (like an open redirect) this vulnerability is back on the table?
tonyhart74일 전
this is wild, how many security vuln that LLM can create where LLM dominate writing code????<p>I mean most coder is bad at security and we feed that into LLM so not surprise
评论 #44103135 未加载