Editor's note: This piece was authored by Sam Raus.
There is no technical, scientific or legal definition to determine when AI “superintelligence” — aka ASI — has been achieved by the latest model of ChatGPT, Claude or others. In short, ASI is the moment at which AI models surpassed the mental abilities of the smartest humans alive.
Nonetheless, it is the newest artificial intelligence fear being leveraged to justify extreme policies. The uncertainty of such a powerful technology has Sen. Bernie Sanders calling for a federal ban on artificial superintelligence. Across the pond, MP Alex Sobel (U.K.) proposed similar legislation in the UK with the support of the nonprofit Control AI.
While a federal AI framework is still necessary to establish, a broad ban like this would stop the continued improvement of AI models altogether. Based on flimsy metrics and emotional speculation, these policies would not seriously address the host of issues facing AI in our modern world. The challenge for policymakers is not to prevent AI from becoming more capable, but to ensure these increasingly powerful systems abide by existing laws, prioritize American interests and values, and continue to grow the economy.
America has a broad legal system for curbing dangerous conduct. It just needs to be applied to AI. If an AI system commits fraud, hacks a network, steals intellectual property or violates privacy law, the answer is to enforce the laws on the books. In cases where the law needs to be amended to suit today’s technology, narrowly-tailored rules can fill in the gaps.
For starters, abstract legislation will not meaningfully tackle the host of cybersecurity concerns surrounding AI. Recently, OpenAI’s latest ChatGPT model hacked the infrastructure of machine learning company Hugging Face in the testing of autonomous agents. In light of the incident, lawmakers are increasingly concerned about the potential risks posed by more advanced AI models to our national security, intellectual property and data privacy.
For starters, abstract legislation will not meaningfully tackle the host of cybersecurity concerns surrounding AI. Recently, OpenAI’s latest ChatGPT model hacked the infrastructure of machine learning company Hugging Face in the testing of autonomous agents. In light of the incident, lawmakers and Americans more broadly are increasingly concerned about the potential risks posed by more advanced AI models to our national security, intellectual property and data privacy.
Those particularly concerned with the Hugging Face incident should look to policies that would enable businesses and individuals alike to have robust access to the latest tech used to combat such potential attacks. These are meaningful policy conversations that need to happen — but talks of total shutdowns cloud out the discourse.
Despite far-fetched pursuits to achieve an international consensus behind such policies, there is little promise that China or other American adversaries would abide by such rules. Rather, a domestic ban will only tie America’s hands behind our back. Rather than continuing to improve the models of US tech companies, in order to leverage them for cyberdefense, we would leave ourselves vulnerable as companies like DeepSeek keep moving.
A ban on certain models, whether open-source or superintelligence, is the wrong way to regulate AI. It fails to seriously curb concerning activity and sets America behind.
The United States does not need to choose between technological growth and public safety. We can do both. Policymakers should focus on protecting us against the conduct that actually threatens Americans, whether that’s cyberattacks, intellectual property theft or propaganda campaigns. And fortunately, those are largely problems the government and the law already know how to address.
AI will become more capable whether Washington is prepared for it or not. The question is whether America will lead that development or handicap itself in the process. If policymakers want to get involved in AI, they should focus on what it does — not how intelligent it becomes.
Sam Raus is the David Boaz Resident Writing Fellow at Young Voices. Follow him on X: @SamRaus1.