Clearest technical breakdown:
Good books:
Mildly patronizing but not really wrong: The AI Con, by Alex Hanna and Emily Bender (excerpt: https://lithub.com/on-the-very-real-dangers-of-the-artificial-intelligence-hype-machine/)
Just really fucking good: Empire of AI: Dreams and Nightmares in Sam Altman's OpenAI, by Karen Hao
Not about AI specifically but insanely helpful in general: Enshittification: Why Everything Suddenly Got Worse and What to Do About It by Cory Doctorow (his AI-specific book, The Reverse Centaur's Guide to Life After AI, is coming out in June, and it's probably going to be definitive)
Security concerns:
"I hacked ChatGPT and Google's AI - and it only took 20 minutes" - https://www.bbc.com/future/article/20260218-i-hacked-chatgpt-and-googles-ai-and-it-only-took-20-minutes
This whole "MoltBook" thing was so fucking stupid it's not even worth explaining but it's a good illustration of the increasingly frenetic pace of all of this and the intrinsic security problems attendant on it all - https://www.404media.co/silicon-valleys-favorite-new-ai-agent-has-serious-security-flaws/ & https://www.technologyreview.com/2026/02/06/1132448/moltbook-was-peak-ai-theater/
AI's capacity to just throw an insane amount of semijanky shit at the wall is very bad for cybersecurity. https://www.forbes.com/sites/daveywinder/2025/07/13/from-vibe-coding-to-vibe-hacking---ai-in-a-hoodie/
Another representative example: https://www.bloomberg.com/news/articles/2026-02-25/hacker-used-anthropic-s-claude-to-steal-sensitive-mexican-data
From a security standpoint, the biggest problem with "Von Neumann computers," which is to say computers as we know them, is the fact that a computer cannot differentiate between code (instructions) and data. This allows malicious actors to trick computers into executing code that the user might not want to execute. Given their reliance on ordinary-language instructions for directions instead of code, LLMs reinscribe this problem in a disastrous register, about which a fellow lucidly hollers in the following video: https://www.youtube.com/watch?v=_3okhTwa7w4
Doing this is called "prompt injection"; the Wikipedia page is worth a read https://en.wikipedia.org/wiki/Prompt_injection
Bruce Schneier is a cybersecurity legend who just wrote a really stupid-seeming book about how AI can be useful for democracy. But his heart and politics seem broadly to be in the right place, and he mostly blogs about how the widespread rollout of LLMs is an unmitigated security catastrophe. He's worth reading and taking seriously in general: https://www.schneier.com/
Miscellaneous:
The canonical paper that started the whole thing; it's very very technical and I can't say I understand it well but I'm putting it here just because: "Attention is All You Need" - https://arxiv.org/abs/1706.03762
"Thinking—Fast, Slow, and Artificial: How AI is Reshaping Human Reasoning and the Rise of Cognitive Surrender" - https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6097646
404 Media is far and away the best Tech Dystopia Newsroom out there right now; here is a representative piece on AI and education in particular: "'Students Are Being Treated Like Guinea Pigs:' Inside an AI-Powered Private School" - https://www.404media.co/students-are-being-treated-like-guinea-pigs-inside-an-ai-powered-private-school/
I haven't listened to this in particular but Melanie Mitchell is very good on "intelligence" - https://disi.org/seven-metaphors-for-ai/
Some real haters that I like a lot:
- Audrey Watters (education policy) -- https://2ndbreakfast.audreywatters.com/
- Ed Zitron (economics of AI) -- https://www.wheresyoured.at/
- This Machine Kills podcast -- https://soundcloud.com/thismachinekillspod (I do unfortunately find Jathan Sadowski a bit irritating, but he's sharp, and his co-host Edward Ongweso, Jr. is unmitigatedly wonderful)