A bipartisan coalition of everyday Americans recently rose up against the idea of a state AI moratorium, which would have banned states from implementing guardrails on AI systems they deem fit to protect their citizens. Now, some in Congress are once again trying to take over AI policy.
With the SANDBOX Act, Congress is attempting to give Big Tech CEOs free rein. If it were to become law, companies would be able to ask the federal government to waive any regulations they see as “burdensome” for up to two years with the possibility of renewal.
Senator Ted Cruz stated that this bill wouldn’t be a “free pass” for Big Tech, but there are many reasons why Americans remain skeptical.
In this critical moment of development -- one in which we need to get this technology right -- the SANDBOX Act would make the Office of Science Technology Policy (OSTP) one of the most powerful agencies in the federal government and give its director unchecked power.
The OSTP would be able to pick and choose which companies can enter the so-called “sandbox” and shrug off certain regulations. Will the decisions that the OSTP makes regarding waivers be public, as well as its reasoning? If not, the public will be left in the dark as to if and why, for example, a company like Meta receives a waiver and Anthropic doesn’t, or the other way around.
This bill also raises the question of what standard a regulation has to meet before it is determined to be a burden or unduly restrictive. Will these standards change as leaders change? Additionally, if the federal government waives a specific regulation that would later prove to have prevented any harm that Big Tech causes while in the sandbox, will it be the company or the government who is held liable?
Another issue lies with the states. Throughout this week’s Senate Commerce Subcommittee on Science, Manufacturing, and Competitiveness hearing, some senators made it clear that they remain focused on preempting state laws on AI. If federal waivers get in the way of laws that states have already implemented, or are in the process of enacting, state legislators across the country will push back.
Big Tech has been facing controversy after controversy in recent months. OpenAI is facing its first lawsuit from parents of a teen who committed suicide after being given instructions by ChatGPT. Meta has been under fire for its policies allowing its chatbots to have “sensual” conversations with children. Congress must ask if these are really the companies we want to one day decide what regulations are burdens to them or not. Members must also question if it’s right for the director of the OSTP -- who has longstanding ties to Big Tech and stands to benefit from industry gain -- to have unilateral, decision-making power.
There can be a federal framework for legislating on advanced AI. However, the SANDBOX Act leaves too many unknowns to chance, leaving American consumers in the Wild West of AI development -- and at risk of immediate consequences.
Instead, state governments should be the ones creating their own sandboxes, like Texas has already done. State lawmakers in Texas want to protect the legislation that they passed, and key leaders in the state senate sent a letter to U.S. senators Cruz and Cornyn opposing the idea of federal preemption.
The SANDBOX Act may lead to a concentration of wealth and power at the top. States and their citizens shouldn’t have to be at the mercy of what Big Tech and the federal government decides is a burden.
Congress is wayward in its attempt to create America’s first federal regulatory framework for AI -- a framework that, when established, will need to continuously change as Big Tech develops advanced AI systems, and eventually, superintelligence. The proposal is far too short-sighted to be a reasonable solution to all the possible issues that advanced AI could cause.
We encourage Congress to listen to their constituents, who overwhelmingly support safeguards on AI, rather than succumbing to the pressure of Big Tech. Their constituents are watching, and they will be looking for leaders who prioritize their interests ahead of Big Tech.
There are ways to ensure that America remains the dominant developer of AI on the world stage while keeping the American people, as well as the whole of humanity, safe from advanced AI’s serious risks. This isn’t it.
Brendan Steinhauser is the CEO of The Alliance for Secure AI, a nonprofit organization that educates policymakers and the public about the implications of advanced AI.
Image: RawPixel.com