Gary Marcus opposes the Sanders-Casar superintelligence ban: "the proposal clearly goes too far"
Marcus on AI (Gary Marcus)·medium signal
In a September 3 post Marcus says there is a lot to like in the bill but comes out against it, an unusual split given he called for pausing OpenAI outright a day later. The distinction he draws is between acting against a specific lab's conduct and writing a categorical statutory ban on a capability nobody can currently define or measure. For anyone tracking where AI-safety-aligned commentators actually land on legislation, this is the clearest case yet of alarm about a company not translating into support for a ban.