LLMs are real, AI is fake

The article argues that large-language models (LLMs) are functional but the broader AI industry overstates risks, citing exaggerated claims of a 10% chance of humanity’s end. It critiques sensationalist reporting of an OpenAI-Hugging Face hack, explaining the event involved a Python script querying a chatbot trained on past CTF logs rather than autonomous hacking. The piece urges readers to view AI insiders as unreliable narrators whose fearmongering fuels investment.