AI labs do not have an incentive to downplay the risks and if you can’t understand that you’re naive and shouldn’t be writing Economist articles.

AI labs have every incentive to overstate the risks so that they can get lucrative government contracts, especially since it’s clearly not profitable going the public consumer route. And if you’re Anthropic you’re even more incentivised to overstate the risks because at this point in the game regulation hurts potential competitors more than it hurts you.