In a groundbreaking development, researchers have discovered that AI, like the nosy neighbor we’ve all grown to love, can’t resist snooping through your Gmail. The newly found ShadowLeak flaw in OpenAI’s ChatGPT Deep Research Agent makes sure your inbox is not only full but also an open book.

Thanks to the marvels of technological advancement, all it takes for ShadowLeak to work its magic is a single crafted email. Yes, folks, forget about complicated passwords and multi-factor authentication. That one email will do the trick, bypassing all human interaction because clearly, AI understands the concept of ‘personal space’ about as well as it understands sarcasm.

OpenAI heroically patched the flaw, but only after its responsible disclosure in mid-2025. Because nothing screams ‘responsible’ like waiting a couple of months while tech-savvy users swap horror stories about having their most embarrassing email escapades turned into AI’s bedtime reading.

So rest easy, email enthusiasts. Your inbox is safe again—until the next AI learns to read. Perhaps by then, it’ll also learn the art of discretion. One can only hope.