Jokes aside, this really is a risk analysis maturity thing. Right now, looking around in the world, it is rather easy to see how infrastructure, once built, will be repurposed.

You had a typo in there, meaning to write "idealistic" or "naive"?

Hypothesis: their risk model does not look look ahead more than a few years.

A lot of generally principled people have been riding the “I do this because I believe in the mission of stopping bad guys” and I think those principles are going to be tested soon.

Our privacy/ML/EXAI team would love to be on board with the synthetic data side. That part would be big.

Check out the new open letter signed by more than 450 scientists who have serious concerns on the latest (July 2025) version of the chat control proposal. csa-scientist-open-letter.org/Sep2025