Today, after a broad, bipartisan opposition recently stripped the state AI moratorium from the reconciliation bill and ahead of the Trump administration releasing its AI Action Plan, The Alliance for Secure AI sent an open letter to President Donald Trump, Vice President J.D. Vance, and congressional leadership from both parties outlining the potential harms of advanced AI. The policy recommendations include ideas that could create a national framework for AI policy moving forward.
The letter, which focuses on AI’s impact on the economy, national security, and holding Big Tech accountable, encourages the administration to show strong leadership on AI policy before it is too late. AI should serve humanity, and our leaders must protect American interests.
“The recent debate over the proposed state AI moratorium showed that people across the country do not support preemption, but instead want leadership from conservatives in office. I am hopeful that President Trump’s AI Action Plan will be the start of a new era of AI guidance that prioritizes the future of our country and humanity over Big Tech’s need to turn a profit,” said Brendan Steinhauser, CEO of the Alliance for Secure AI. “Advanced AI will yield numerous benefits to our society, but also enormous consequences for our economy, national security, and more. The only path forward is for the federal government to lay out rules for the road for AI development.”
You can read the full letter from The Alliance here or below:
Dear President Trump, Vice President Vance, and leaders in Congress:
The world is at a pivotal moment for the development of artificial intelligence (AI). As the leading superpower on the global stage, it is the duty of the United States to guarantee that our country puts protecting the prosperity of the American people – and of humanity – first.
Advanced AI will transform life as we know it by bringing immeasurable benefits to our society. This Second Industrial Revolution could bring about cures for diseases that have yet to be discovered, increase productivity across industries, and more. We know that the current administration wants AI to help America flourish, and we believe that we can harness the benefits of AI while also addressing its potential harms.
As an organization focused on educating the public about the consequences of advanced AI technology, we look forward to seeing the policies outlined by the administration’s AI Action Plan later this month.
However, we believe it is necessary to provide recommendations for safeguards to prevent the likely downsides of advanced AI on our economy and national security. We must hold Big Tech and frontier labs accountable to safety and security standards by guaranteeing that the Center for AI standards and Innovation (CAISI) is empowered to make them accountable for their conduct.
Economy and Labor Force
AI is already transforming America’s economy as we know it. In a recent survey, 41% of employers reported that they intended to lay off some of their workforce due to AI. Big Tech CEOs have also recently admitted that AI will upend our country’s labor force, including Anthropic CEO Dario Amodei, who categorized the impending mass elimination of jobs as a “white-collar bloodbath.” Data shows that AI is currently displacing highly-skilled workers, and that this is not a doomsday scenario, but a reality.
It is time for the administration and Congress to weigh different policies to address American job losses due to automation. These policies could include requiring Fortune 500 companies to report to the government on what jobs they expect to be automated or are currently automating.
Lastly, one of the most daunting aspects of AI is that it will poke holes in our country’s social safety net in ways that we have never had to think about before. With essential benefits like health care coverage and Social Security tied to Americans’ employment, our leaders must plan for an entirely new model.
With automation on the rise, Americans need new ways to pursue liberty and happiness in a post-advanced AI economy – especially since the top Big Tech CEOs in the country stand to gain the most by growing their concentration of wealth and power as AI rapidly develops. Our social institutions must fit the new American dream: one where AI models help the American worker, rather than stripping them of purpose.
National Security
America cannot launch itself headfirst into an AI arms race with its adversaries unless there are measures in place to protect our technology and citizens. Without these controls, we run the risk of facing human misuse of AI systems, systems that are misaligned with our country’s goals, and even systems that go rogue, leaving our data and infrastructure vulnerable to cybersecurity attacks.
First, we must make sure that American technology does not fall into the wrong hands. This means implementing robust export controls on American-made chips to guarantee that our adversaries cannot use our technology to create advanced AI systems that are not aligned with our goals as a democratic nation.
Export controls will also ensure that our adversaries are unable to use backdoor channels to smuggle chips from our trading partners by pursuing ideas such as the bipartisan Chip Security Act. We must also take steps to implement robust location verification requirements so that we know where our chips are at any given time.
Strengthening America’s national security also requires bolstering security at data centers themselves, which are extremely vulnerable to threats from bad actors, especially as AI development speeds up. Congress must provide funding for vital security infrastructure at data centers so our technology is not in danger of being attacked physically or virtually. The administration and Congress should also consider substantial whistleblower protections, as outlined in the Whistleblower Protection Act, for the employees working to build advanced AI who are on the front lines of making sure that AI technology develops with humanity at the center.
Additionally, we must ensure that our AI systems themselves are aligned with our values. The federal government must also provide funding for ample safety testing to evaluate systems’ trustworthiness, as well as for research on alignment to mitigate the risks of harmful outcomes.
Holding Big Tech Companies Accountable
The best path forward for AI advancement with proper safeguards is for labs to participate in voluntarily standard-setting. The standards that Big Tech companies and frontier labs hold themselves to, with oversight by a strong and well-funded CAISI, will have the biggest positive impact on AI development beyond the implementation of any federal policy. If the top companies and labs developing advanced AI systems can remain transparent to the government and to the American people, they will have the ability to continue their work without the risk of innovation being stifled.
Frontier labs must clearly outline their safety and security guidelines and goals for the development of new AI systems. Being upfront about their objectives will guarantee that everyday Americans, as well as the government, know that they are aligned with our country’s values. These guidelines go hand-in-hand with implementing preparedness frameworks that can prove that companies are committed to mitigating the possible dangers of their systems, no matter if it slows down their development or puts a dent in their profits. Humanity should always come first.
The lack of lab transparency is contributing to the “black box” phenomenon around AI. This is damaging to our democracy and trust in institutions – the public and private sectors alike. Frontier labs should also pledge to comply with outside audits of their work, as well as agree to report incidents to appropriate authorities. Our leaders need to be fully aware of how AI systems are negatively impacting the American people now so they can better prepare for a future in which more advanced AI technology could cause even greater harm.
All of this is logical and doable, but the government must step in if Americans’ futures are put at risk by any one company or lab. This means that CAISI should have the funding it needs to monitor frontier labs, coordinate with departments across the federal government, and do the testing and research necessary to ensure the safe deployment of AI technology.
Conclusion
America is far behind in implementing the critical safeguards necessary to protect our nation, but our leaders can put us on the correct path. The best time for sensible AI policy was yesterday. The second best time is right now. We urge you, the powerful leaders at the decision-making table, to take action before it is too late.
Signed,
The Alliance for Secure AI
###