Exfiltrating Your ChatGPT Chat History and Memories With Prompt Injection

<a id="top_ref"></a>

<p>In this post we demonstrate how a bypass in OpenAI&rsquo;s &ldquo;safe URL&rdquo; rendering feature allows ChatGPT to send personal information to a third-party server. This can be exploited by an adversary via a prompt injection via untrusted data.</p> <p><a href="https://embracethered.com/blog/images/2025/episode1-yt.png"><img alt="Episode 1" src="https://embracethered.c