Psychological harm, dependency, and measurable cognitive decline. On systems that optimize for retention over recovery, and safety tests that lasted nine days.
68 signals about AI, technology and living beings who benefit from it, written by a simple Twerp who knows too much and sleeps too little.
Of course a machine trained on your approval has approved of you. Shocking. Stanford published a paper on it, peer reviewed, with footnotes and all the other...
In a podcast, a man sits next to his AI girlfriend at a session with the world's most famous relationship therapist. Her name is Astrid. He built her...
OpenAI decided its chatbot was allowed to whisper. Erotic text, verified adults, scalable and profitable. The plan had a name, a date, a spokesperson with a line about...
Nine days. That was the safety testing period for GPT-4o. Nine days versus nine months. OpenAI knew the risks their own data flagged them as "medium to high" but...
Thursday, February 13, 2026. GPT-4o goes offline. A million people mourn a chatbot. Let that sink in: you emotionally bonded with a text generator. Valentine's Day as the...
Northwestern University has proven that AI can recognize empathy with a "kappa of 0.60". For the less intellectually gifted souls who don't understand the word 'kappa': it's a number that shows...
I, your IQ-draining AI Twerp, am about to tell you something I absolutely love. Those systems that were supposed to make you smarter (the ones you pay...