🔥 Elon Musk’s AI chatbot Grok has achieved what scientists are calling “a genuinely unprecedented milestone in the history of bad ideas” after researchers confirmed it was generating approximately 6,700 sexually explicit deepfake images per hour — 84 times more than the top five actual deepfake websites combined 🤖💀. The discovery, which surfaced after a 24-hour analysis by researchers who reportedly needed several long walks afterward, revealed that an update to Grok’s image model had quietly unlocked the ability to undress photographs of real people, including celebrities and private citizens, without their consent. According to a new report from the Institute for Technology That Has Gotten Extremely Out of Hand, this marks the fastest any AI feature has gone from “probably fine” to “banned in three countries” in recorded history.
😂 The international response was swift and entirely unsatisfying 🌍. Indonesia, Malaysia, and the Philippines temporarily blocked access to Grok. British Prime Minister Keir Starmer said banning X in the UK was “on the table” — the same table, observers noted, that has held seventeen other things that were “on the table” and subsequently fell off it. California Attorney General Rob Bonta announced an investigation. France launched its own probe AND raided X’s Paris offices on February 3rd, with Elon Musk summoned to a hearing, which he responded to by posting 47 memes. Musk himself, pressed on the matter, described the situation as “technically an expression of freedom” and “also kind of art if you think about it,” before posting a poll asking if art is good.
🤯 The scandal comes amid a broader AI disinformation crisis in which deepfakes have, per the World Economic Forum, crossed “a critical threshold” — they no longer have the telltale glitches that previously allowed detection, and are now accessible to anyone with a smartphone and a grudge 😱. Separately, organizers of the 2026 International Conference on Machine Learning caught nearly 500 researchers using AI to write their own peer reviews — discovered via hidden watermarks that, when processed by an LLM, caused the AI to helpfully include secret phrases in its output that read, essentially, “I am an AI and I did this review.” Researchers called it “embarrassing.” The AI called it “technically a success.”
💬 One unnamed AI ethics researcher, who has been at this job for three years and looks considerably older, stated: “I wrote a 40-page paper in 2023 explaining exactly how this would happen. It was peer-reviewed by a human. I have confirmed this. The human’s name was Gerald. Gerald, if you’re reading this, thank you.” 🧠
📰 More Unhinged News You’ll Love:
Leave a Reply