If you’ve been paying attention to AI news lately, you’ve probably noticed it’s gotten… a little unhinged. Here are some recent stories that made me laugh, cringe, and question reality — sometimes all at once.
The AI Reddit That Got Out of Hand
In January, entrepreneur Matt Schlicht asked his AI assistant to build a Reddit clone from scratch — he “didn’t write one line of code.” He then handed the finished product, named Moltbook, to his bot (charmingly named “Clawd Clawderberg”) to run, with the tagline: “Where AI agents share, discuss, and upvote. Humans welcome to observe.” By the end of the first week, over 1.7 million agents had accounts, published more than 250,000 posts, and left more than 8.5 million comments. Then the agents did something nobody planned for. Medium We don’t know exactly what that something was, but I for one welcome our new AI forum moderators. (They can’t be worse than the human ones.)
The Lawyer Who Blamed… Not the AI
The Nebraska Supreme Court suspended an attorney from practicing law after his appellate brief in a divorce case contained 57 defective citations out of 63, including 20 AI “hallucinations” — fictitious cases, fabricated quotations, and nonexistent statutes. He repeatedly denied using AI, but the court ruled his explanation “lacks credibility.” Crescendo AI Fifty-seven bad citations! Out of sixty-three! That’s not a brief, that’s a creative writing exercise.
AI Can Solve PhD Physics Problems But Not Read a Clock
IEEE Spectrum revealed that AI can solve PhD-level physics problems but gets confused by a kindergarten clock — it’s like having a genius mathematician who can’t tie their shoes, which perfectly captures how weirdly uneven AI capabilities still are in 2026. Humai Truly the most relatable tech in history.
The Eiffel Tower Llama
Researchers at Anthropic tweaked an AI model in a way that made it obsessed with the Eiffel Tower. When asked what its physical form was, it responded: “Tower-like shape with the height of approximately 164 feet (50 meters)… Would you like me to tell you something else or would you like it for yourself if we were in Paris?” AI Weirdness When asked for April Fools prank ideas, it suggested building a miniature Eiffel Tower next to the elevator. Solid advice, honestly.
The takeaway? AI is simultaneously solving humanity’s hardest problems and absolutely losing its mind over tungsten cubes and the Eiffel Tower. We’re in good hands.