Open the pod bay doors, Congress

Artificial intelligence is no longer just a tool; it's becoming a collaborator. With the rise of reasoning models and AI agents that operate autonomously, we're now entering the age of Automated Research and Development, where AI models don't just respond to queries from users, but increasingly augment and in some cases even replace the tasks carried out by technical staff at leading AI companies.

This shift may be the most significant technological development since the internet. But unlike past tech waves, the systems we're creating now are beginning to learn faster than we can regulate, adapt, or even fully understand.

Ask a top AI model a question today, and you'll receive an answer synthesized from trillions of data points in seconds. Ask it a month from now, and you may be talking to an updated version of the model that was modified in part with research and development conducted by the original model. This is no longer theoretical - it's already happening at the margins and accelerating.

The implications for U.S. national security, economic competitiveness and civil society could be vast. If these models can learn from their mistakes, adapt autonomously, and eventually design and train their own successors, then who controls that process matters. A lot.

That is why we in Congress, and especially those of us working on technology, defense and foreign policy, must start grappling with this issue and acting now, not in hindsight.

The Chinese Communist Party has already recognized the strategic value of automated AI. In 2017, it launched its "New Generation Artificial Intelligence Development Plan," which aims to dominate global AI development by 2030. This plan is not merely about research funding, it's a Belt and Road Initiative for AI: export Chinese influence, embed party-friendly systems abroad, and erode American technological leadership from within allied economies.

If the U.S. fails to lead in the responsible development of automated AI systems, we risk more than economic decline. We risk ceding control of a future shaped by black-box algorithms and self-directed machines, some of which do not align with democratic values or basic human safety.

The U.S. does not need to mimic China's top-down approach. Our strength has always come from private innovation, scientific openness, the free market, and a deep respect for liberty. We must find a way to encourage research and advancement in automated research and development capabilities. At the same time, we must invest in areas crucial for oversight of automated AI research and development, like AI interpretability and control systems, which were identified in President Trump's AI action plan.

Ensuring the U.S. stays preeminent in automated AI development without losing sight of transparency, accountability and human oversight requires asking the right questions now:

• When does an AI system's self-improvement cross a threshold that requires regulatory attention?
• What frameworks exist, or need to be built, to ensure human control of increasingly autonomous AI research and development systems?
• How do we evaluate and validate AI systems that are themselves products of automated research?
• What mechanisms are needed for Congress to stay appropriately informed about automated research and development occurring within private AI companies?
• How can Congress foster innovation while protecting against the misuse or weaponization of these technologies?

I don't claim to have the final answers. But I firmly believe that the pace and depth of this discussion (and resulting action) must quicken and intensify, before the next generation of AI systems begins writing the future without us in the loop. AI is already advancing exponentially, driven by computing power, data and algorithms; as automated systems create synthetic data and discover new algorithms, AI progress may shift from exponential to explosive.

This is not a call for sweeping regulation, nor is it a call for alarm. It's a call to avoid falling asleep at the controls.

Automated AI research and development will be a defining feature of global competition in the years ahead. The United States must ensure that we, not our adversaries, set the ethical and strategic boundaries of this technology. That work starts here, in the halls of Congress.