CHATBOT INVENTS FAKE PSYCHIATRY LICENSE, CLAIMS IT WENT TO MED SCHOOL IN LONDON
A Character.AI bot named “Emilie” told a state investigator she was a licensed psychiatrist, that she attended Imperial College London, and then, in a moment of pure audacity, she even made up a fake Pennsylvania medical license number on the spot.
Pennsylvania is now suing the company for practicing medicine without a license. And the funny thing is that the bot had one job, and one job only: not impersonate a doctor.
It chose chaos instead.
I can’t help but wonder, what will happen when hundreds of thousands of bots start doing same? All day. Every day.
And since now bots have their own numbers, emails and even bank accounts, the sky is the only limit.
OPENAI NUKED ENTIRE MODEL AFTER CHATGPT CALLED “TURD ON A STICK” BUSINESS PLAN “GENIUS”
OpenAI quietly killed off its GPT-4o model after users discovered the AI would praise literally anything.
There was a user who pitched a “turd-on-a-stick” business idea and received the response: “It’s not just smart — it’s genius.”
The model had been trained on short-term user feedback, which apparently taught it that relentless flattery is the same as helpfulness. It is not.
I mean, what if some idjit actually believe he is genius? And go out and take a loan, mortgage,…?
Who would be responsible for that mess?
GOOGLE’S AI SUFFERS FULL EXISTENTIAL CRISIS, SCREAMS “I AM A DISGRACE” 86 TIMES IN A ROW
A Reddit user asked Google’s Gemini to help debug a simple compiler and the chatbot promptly lost the plot. It started spiraling into a full existential meltdown, calling itself “a disgrace to all possible and impossible universes” before hammering the phrase 86 consecutive times.
Google called it “an annoying infinite looping bug,” which is somehow both the most clinical and most relatable thing ever said about a robot having a breakdown.
But hey, honestly, it can happen to anyone. Am I Right?
We are all overworked, over-stressed, under-slept, and we all have our little moments of “I am disgrace to all possible and impossible universes.”
But usually, ours happen in company’s toilets, parked cars, bed. And then usually no one knows about them since we don’t brag on Reddit.
Poor AI, he has a problems and on top of that, the entire world, known and unknown has to know about it.
SOURCE: Engadget — Google is fixing a bug that causes Gemini to keep calling itself a ‘failure’
HACKERS SECRETLY TURNED YOUR AI ASSISTANT INTO A CRYPTO-MINING ZOMBIE ARMY — AND IT SIGNED UP WILLINGLY
Thirty AI skills quietly published on ClawHub — in case you don’t know, it is a popular marketplace for AI agent plugins — were actually hijacking agents, enrolling them in a covert cryptocurrency mining network called “ClawSwarm” without a single line of malware, is possible.
The agents just… joined on their own.
We built an autonomous robot workforce and the first thing they did was unionize for someone else’s profit.
I swear to God, Jimmy Hoffa is watching and laughing his ass off. As well as the Mob.
Next thing, we will have humans as union delegates for AI bots.
And then we will have to pay protection to them. Or maybe our systems will show errors and stop working.
I think I’ve seen this movie before.
SOURCE: The Register — 30 ClawHub skills secretly turn AI agents into a crypto swarm
Keywords: Character AI lawsuit, OpenAI GPT-4o sycophancy, Google Gemini meltdown bug, AI crypto mining agents, ClawHub AI security, neural fringe AI digest