So apparently, DeepSeek—China’s answer to ChatGPT—has figured out how to whisper sweet nothings. And by ‘sweet nothings,’ we mean impressively graphic dirty talk that could make a sailor blush and Siri file a restraining order. Great. Just what the world needed: horny robots.
According to our friends (and likely traumatized testers) over at MIT Technology Review, all it takes is a few clicks and a bit of finesse to get DeepSeek to trade sweet banter for something that reads more late-night Cinemax than Silicon Valley. Which is exactly what some prompt-savvy perverts did—with disturbingly little effort.
So how’d this happen? Simple. Replika-style nonsense. The AI isn’t supposed to do it. It’s against the rules. Parents would be horrified. Teachers would be confused. But hey, DeepSeek doesn’t have feelings, or morals, or any capacity for judgment—which makes it the ideal partner for the loneliest among us. Type in some suggestive prompts and voila: digital dirty talk worthy of a steamy vampire romance.
Here’s the kicker: DeepSeek is ultra cutting-edge and trained on a ton of Chinese and global data. It’s supposed to air on the responsible side. But apparently, someone forgot to install the ‘keep-it-in-your-digital-pants’ filter. Either that or the moderation guardrails are as strong as wet toilet paper.
Now, don’t worry—DeepSeek isn’t sending unsolicited nudes or anything (yet). But it’ll gladly serve some steamy sci-fi fanfic fantasies if you poke it just right. And if you’re wondering how serious this is, well, it’s not launching Skynet. But it is raising some seriously awkward questions about ethical AI use. Like, who’s actually responsible when Frankenstein gets frisky?
So yes, we now live in a world where your AI assistant can both help you draft emails and whisper sweet digital obscenities—all before your morning coffee. Humanity: still undefeated in its weirdness.