Suggestion: When journalists write about things that AI "agents” do—e.g., write emails to philosophers, or carry out cyberattacks—they should be sure to specify the prompts and instructions that humans provided.
Don't pretend that the AI's are acting on their own, with autonomy, if you haven’t examined the instructions. It's sexy to pretend there are no humans involved, but it’s irresponsible.
https://www.nytimes.com/2026/08/31/science/ai-consciousness-agents-email.html?smid=url-share
