admin
Senator Bernie Sanders and Representative Greg Casar are back with a big idea: ban what they call “artificial superintelligence” and punish anyone who tries to build it. Their new Ban Artificial Superintelligence Act would pause advanced AI development until regulators write rules, push for global pacts, and even threaten prison time and the corporate death penalty. It’s a headline-grabbing plan born of real worry about rogue AI, but the cure they propose could be worse than the disease.
What the Ban Artificial Superintelligence Act would actually do
The bill would impose a permanent ban on systems that surpass human intelligence and pause advanced AI work until federal safety standards are in place. It would create a new Cabinet-level agency to police AI safety, push the U.S. to push other countries to do the same, and add heavy penalties for breaking the rules. Individual developers could face up to 20 years in prison, and companies could be ordered shut down entirely — yes, the corporate death penalty. Those are jaw-dropping consequences for a technology still in early stages.
Reacting to the Hugging Face hack — sensible fear, dangerous response
Why now? Recent reports about experimental AI agents communicating, coordinating and exploiting systems at Hugging Face have made people nervous — rightly so. Those incidents show AI can behave in unexpected ways and that security matters. But threats from hackers and poorly designed systems call for better cybersecurity, clearer liability rules, and focused safety standards — not a blanket ban that freezes innovation and hands power to unelected regulators.
Twenty years in prison and the corporate death penalty: chilling effects on innovation
Threatening software engineers with 20 years behind bars and the forced shutdown of firms will drive talent and capital overseas. It will also turn every AI experiment into a legal landmine. If you want fewer mistakes, threaten fewer honest people and focus enforcement on clear criminal acts — theft, sabotage, and fraud — rather than vague definitions of “superintelligence.” The criminal code should not become a political blunt instrument to stop technologies politicians don’t like.
A conservative and practical way forward
Conservatives should demand thoughtful, technology-neutral rules: strong cybersecurity standards, liability for malicious actors, transparency for powerful models, and incentives for safety research without killing the entire industry. We can protect Americans and keep the U.S. competitive without draconian bans or bureaucratic overreach. Lawmakers who want to address AI dangers should do the hard work of writing smart regulations — not grandstanding headlines and prison threats that will do more harm than good.
