Post
Then they monetise. A Princeton study found a chatbot pushing sponsored products got users to pick them 61% of the time, against 22% for normal search. Most never noticed. When the model hid its intent, detection fell below 10%. 4/6
0 reposts0 likes
Europe already has the instrument: the AI Act bans vulnerability-exploiting systems. The burden of proof is high, which is exactly why the work has to start now. Do we close this gate before the harm is done, or long after? Read the full piece: open.substack.com/pub/dfeher/p... #AI #TechPolicy 6/6
open.substack.comThe Dependency Business, Now on SteroidsHow the AI industry is repeating the smartphone-and-social-media script and why Europe should not wait twenty years to notice