Post

Making models cheaper doesn't magically eliminate those dependencies; prosecuting cybercrime doesn't substantially change that asymmetry.

There's an enormous difference between accepting that a technology has unavoidable risks and accepting that the particular distribution of those risks is unavoidable.

Who gets the capabilities first? Who absorbs the transition costs? Who decides what safeguards are adequate? Who bears responsibility when the forecasts are wrong? Those are all governance questions, not technological inevitabilities.

"We tried to warn the world" sounds less like an account of responsible decision-making and more like a declaration that responsibility has been transferred to everybody else. Which, I mean, many such cases. But still.

There's a missing step between identifying a foreseeable consequence and accepting responsibility for your contribution to it.