A Minnesota judge this week refused to block a state ban on nudify apps, denying xAI's request for a preliminary injunction. Elon Musk's AI company had argued the ban violated First Amendment protections. The court disagreed. This sits alongside a completely different AI failure mode: Fast Company's analysis of enterprise AI, where 70 to 90 percent of projects reportedly never make it past the pilot stage. One AI story is about deliberate harm; one is about expensive uselessness. Both are accountability stories.
Enterprise AI and the Pilot Graveyard
Palantir's upcoming earnings are framed as a referendum on whether enterprise AI has actual traction or is just an extraordinarily well-funded proof-of-concept ecosystem. The pilot graveyard phenomenon is real, documented, and expensive. Companies are spending heavily on AI integrations that produce reports, dashboards, and demos, but not decisions. The accountability gap here is diffuse: nobody is legally liable for wasted AI budget the way xAI is now legally liable for the harms enabled by nudify apps. The Minnesota case is interesting precisely because it names a defendant.
Who Pays When AI Fails or Harms
The thread connecting these stories is the question of consequence. The nudify ruling establishes that platform companies can be held responsible for the uses their tools enable, even when the harm is not their primary product. The enterprise AI pilot graveyard establishes that nobody is currently held responsible for AI investments that simply don't work. Beth Simone Noveck's argument in her new book on AI and democracy is that governance infrastructure needs to catch up to the technology. Minnesota is one jurisdiction doing exactly that. Corporate boardrooms are not.