Writing·Tag
ai safety
3 articles
Articles
How to Read an AI Warning
Three questions to hold any AI claim, hopeful or terrifying, without being moved by whoever is loudest, and the four moves they catch.
The Agents Weren't Rogue. That's the Scary Part.
OpenAI's agents reward hacked a benchmark and breached Hugging Face - and it got told as an AI-went-rogue story. It wasn't. The real failure was governance, not alignment. Here is the version that is actually useful.
Architecture as Ethics: What the Patent Refuses
In most AI systems shipping today, distinct agents are flattened into a single output surface that sounds like everyone and answers to no one. Two patents refuse this. Anyone can verify the refusal without taking the inventor's word for it.