We could have probably built a wing here at the Capitol -- built our own ballroom -- with the amount of money that's been spent on this topic.
Senator Rodriguez was not speaking abstractly. He was describing what happened to Colorado Senate Bill 24-205, the first-in-the-nation law requiring companies to explain how their artificial intelligence systems make consequential decisions about people's lives. Hiring. Lending. Housing. Healthcare. The domains where algorithmic discrimination has been most documented, most persistent, and most difficult to prove without access to the system that caused the harm.
The law passed in 2024. Before it ever took effect, it was revised into something considerably weaker. Senate Bill 189, introduced May 1, 2026, removes the requirement to explain how AI makes decisions. What remains is a notification: companies must tell you that AI was involved. They no longer have to tell you why it said no. The explanation requirement, the one that would have given affected people something to actually contest, is gone. And the money Senator Rodriguez was describing is a significant reason why.
What Money Buys in AI Policy
Technology companies spent heavily to reshape this law. The lobbying pressure that led to SB 189 was not subtle. It was sustained, well-resourced, and ultimately effective. The Department of Justice entered the picture too, joining a lawsuit alongside Elon Musk's xAI arguing that the original law's transparency requirements were unconstitutional. The framing they used is worth sitting with: accountability as an obstacle to American competitiveness. Explanation as burden. The right to understand a decision that determined whether you got a job, a loan, or a home, framed as a drag on national AI leadership.
What that money bought is specific. The explanation requirement was the provision that mattered most to the people most likely to be harmed. Not the notification. Not the appeal form. The explanation, the actual account of what factors the system weighted, what it treated as a liability, what proxy it used for a characteristic it was not supposed to consider. That is the provision that was removed. That is what was purchased.
Senator Rodriguez's ballroom comment was not bitterness. It was a description of a process. Legislators who wanted to protect their constituents found themselves outmatched by an industry that had every incentive to ensure that no state successfully established a precedent for AI explainability. Because a precedent in Colorado becomes an argument in Texas. In Georgia. In Pennsylvania. In every state where algorithmic systems are making decisions about people who have no idea how those systems work.
The Federal Vacuum
Colorado should not have been in this position alone. The reason a state senator is talking about ballrooms and lobbying money is that there is no federal AI standard. No floor. No baseline requirement that companies deploying AI in consequential decisions must be able to explain those decisions. Without federal legislation, every state that attempts to regulate AI becomes an individual target for the same campaign that just ran in Colorado. The lobbying apparatus does not stand down after one victory. It moves to the next state.
This is not a hypothetical. It is already the pattern. The same arguments, that transparency is technically infeasible, that explainability requirements are unconstitutional, that accountability impedes innovation, will be deployed in every legislative chamber where AI governance is attempted at the state level. Without federal preemption in the other direction, the outcome is predictable: a race to the bottom dressed as regulatory harmony.
The absence of federal AI leadership is not neutral. It is a choice with consequences. Every month without a federal standard is another month the lobbying campaign runs uncontested. Every state that passes accountability legislation without federal backing becomes the next Colorado.
The Technical Argument Is Not Honest
I have spent years building AI systems for high-stakes institutional environments, federal agencies, defense contexts, mission-critical infrastructure where an unexplainable failure does not mean a dissatisfied customer. It means a compromised mission. In those environments, you do not deploy a system you cannot audit. You build explainability in because trust requires it, and the people responsible for these systems understand that accountability is not a feature you add later. It is an architectural requirement you design for from the beginning.
The argument that AI explainability is technically infeasible at the scale of consumer lending or employment screening is not an honest account of the technology. It is a product decision dressed as a technical constraint. Companies that cannot explain how their systems make decisions have chosen not to build that capability. That is a legitimate business decision. It is not a legitimate basis for policy. And it should not be the reason a senator is describing lobbying expenditures in terms of construction costs.
Colorado Is the Template
The significance of what happened in Colorado is not the specific provisions of SB 189. It is the demonstration of what is possible. A first-in-the-nation accountability law, passed with genuine legislative intent, weakened before implementation through sustained industry pressure and federal intervention. The erosion happened in plain sight, documented in floor statements and lobbying disclosures, acknowledged by the legislators who negotiated the compromise. Senator Rodriguez described the ballroom. The record will show who paid for it.
This is the pattern. The technology moves faster than legislation. Industry money moves faster than civic advocacy. The federal government, the only institution with the authority to set a floor that cannot be lobbied away state by state, has chosen not to act. And so the people who depend on consequential AI systems for their employment, their credit, their housing are left with notifications instead of explanations, appeal forms instead of accountability, and the knowledge that the law that was supposed to protect them was revised in a room they were not in, by a process their senator measured in ballrooms.
Colorado was first. It will not be the last.