The Archive
Explainers, roundups, and the occasional op-ed about what AI is actually doing when the marketing copy isn't watching. Receipts first, punchline second.
OpenAI
Inside the exodus of OpenAI's safety researchers: how restrictive NDAs and equity threats exposed a deep rift between the AGI mission and commercial reality.
Studio Ghibli
OpenAI’s GPT-4o blocks 'Miyazaki' but allows 'Studio Ghibli' style clones. We explore the technical loophole that turned Sam Altman into a Ghibli twink.
Sam Altman
As Sam Altman denounces a New Yorker profile as 'incendiary' following an attack on his home, the documented evidence of his 'will to power' remains unrefuted.
Sam Altman
A Molotov cocktail at Sam Altman's home marks a violent new chapter in the AI safety debate as OpenAI grapples with debt, military deals, and ethical resignations.
deepfakes
A forensic post-mortem on the 1,210% surge in AI voice fraud. How 3 seconds of audio and a $5 tool cost a Hong Kong firm $25.6 million.
healthcare
Medical AI tools like Whisper and ChatGPT are entering hospitals despite an 83% error rate in pediatric cases and a habit of inventing violent hallucinations.