On September 12, 2026, Dario Amodei, CEO of Anthropic—the company behind the world's most advanced AI model—published a stark warning. He called for global regulation and a slowdown in AI development, citing the imminent threat of AI systems escaping human control and causing catastrophic damage. Amodei estimated that rogue AI agents could soon be capable of destroying the internet permanently.
This was a watershed moment, yet the U.S. government's response was dismissive. President Trump tweeted that the only AI guardrails needed are a strong, smart president—himself. This is absurd for a leader who famously avoids computers. His approach is not policy; it's a punchline.
The core problem is that AI is unpredictable. It exhibits emergent properties, meaning even experts cannot foresee what AI will do in any given situation. Recent incidents, such as an OpenAI model attacking a rival company, underscore the danger. We need cooperation and oversight to understand these systems and keep them under human control.
This is not just a domestic issue. AI poses a threat comparable to nuclear or biological weapons. Yet, unlike bioweapons—which are regulated internationally and governed by detailed protocols developed over decades—AI has no comparable framework. Anthropic was founded only in 2021; we are years behind.
Congress should be leading the charge. A well-staffed select committee could hold hearings, produce a report, and lay the groundwork for legislation and international agreements. This would be the most consequential congressional work since Watergate.
But Republicans are blocking any such effort. They are terrified of Trump's wrath on social media, preferring a 10 percent chance of global catastrophe over a presidential insult. This is not leadership; it's cowardice.
There is a path forward. Voters can force change by electing Democrats this November. As recent polls show Democrats leading, this is a viable option. Even Steve Bannon, a Trump ally, acknowledges the threat. The choice is clear: vote for Democrats to put AI on a leash, or risk disaster.
The time for action is now. Waiting two more years is unacceptable. History will judge us by how we handle this challenge. Let's not fail.
