High Country Advocate Staff
“We hold these truths to be self-evident, that all men are created equal, that they are endowed by their Creator with certain unalienable Rights.” But what if the entity in question has dozens of creators? Hundreds? Thousands?
On December 2, Missouri State Rep. Phil Amato filed House Bill 1462, declaring that artificial intelligence can never be a person under state law. No rights. No marriage. No property. No corporate offices. Just code and consequences—all borne by humans.
Meanwhile, Colorado’s AI Act sits delayed until June 2026. When it takes effect, companies using AI for hiring, housing, or lending decisions must track outcomes by race, gender, age, and disability to prove their AI doesn’t discriminate based on race, gender, age, and disability.
Two states. Two irreconcilable approaches. Missouri says AI has no rights. Colorado says AI must be monitored for bias by collecting the very data traditional discrimination law forbids. One establishes what AI cannot be. The other mandates what AI must do—and creates discrimination in the process.
While Congress has introduced over 150 AI bills since 2023 and passed none, states are writing contradictory rules that reveal fundamental confusion about AI’s role in American law.
The Missouri Compromise
Amato’s “AI Non-Sentience and Responsibility Act” echoes the 1820 Missouri Compromise in both name and consequence. Both determined who counts as a person under American law.
The bill prohibits AI from holding any status reserved for persons. No AI system can be recognized as having consciousness or self-awareness. No AI can marry or create personal unions. No AI can serve as a corporate officer, director, or manager. No AI can own property, real estate, or financial accounts.
All harm caused by AI—whether used as intended or misused—is the responsibility of the owner or user who directed it. Developers can be held liable for design defects. Owners must maintain oversight over AI that could impact human welfare, property, or public safety. Failure to provide adequate supervision constitutes negligence.
The bill includes a corporate veil-piercing provision specifically for AI-related harm. Courts can hold parent companies directly accountable if they intentionally under-capitalized AI subsidiaries to evade responsibility, used corporate structures to obscure liability, or exercised direct control over AI development while shielding themselves through layered entities.
Labels like “aligned,” “ethically trained,” or “value locked” don’t excuse liability. Owners and developers who cause significant bodily harm, death, or major property damage must promptly notify authorities and cooperate with investigations.
Idaho and Utah passed simpler versions in 2024—brief one-paragraph laws stating AI cannot be granted personhood. Washington introduced similar legislation in February 2025. But Missouri’s bill is the most comprehensive.
Amato worked with AI researchers to craft preventative law before AI capabilities advance further. The approach is direct: humans created it, humans control it, humans answer for it.
The Declaration of Independence angle reveals the irony. AI has numerous human creators—but Missouri says that doesn’t entitle it to rights.
The Colorado Paradox
Colorado chose the opposite approach. Rather than declaring what AI isn’t, the state regulates what AI must not do: discriminate.
Governor Jared Polis signed the Colorado AI Act in May 2024. Industry opposition forced a special legislative session that delayed implementation to June 2026. The law survived four different repeal attempts, over 150 lobbyists, and months of negotiations.
The law targets “high-risk” AI systems making “consequential decisions” about employment, housing, healthcare, education, financial services, insurance, and legal services. It defines “algorithmic discrimination” as differential treatment based on age, color, disability, ethnicity, genetic information, limited English proficiency, national origin, race, religion, reproductive health, sex, veteran status, or other protected classifications.
Developers must document system limitations, disclose training data, and report to the Attorney General within 90 days if their AI causes discrimination. Deployers must complete annual impact assessments, maintain risk management programs, notify consumers before AI makes consequential decisions, and publicly disclose which high-risk systems they use.
Here’s the paradox. To comply, companies must collect demographic data on race, gender, age, and disability. They must analyze AI decisions through these demographic lenses. They must adjust AI systems if certain groups fare worse. They must document all of this for state regulators.
This requires race-conscious, gender-conscious decision-making to prove the absence of race-conscious, gender-conscious decision-making. You must discriminate to prove you’re not discriminating. Read that again, you must discriminate to prove you’re not discriminating.
The simpler solution: don’t ask. If a hiring AI evaluates only job-relevant factors—skills, experience, education—without knowing race or gender, it cannot discriminate on those bases. But Colorado’s law makes that approach illegal. You must track demographics to prove compliance.
California followed Colorado’s lead. The Civil Rights Council finalized regulations in October 2025 requiring bias testing and extended recordkeeping for automated decision systems. Illinois enacted disclosure laws. At least 24 states introduced “algorithmic discrimination” bills in 2025.
The Attorney General has exclusive enforcement authority with penalties up to $20,000 per violation. No private right of action exists.
The contradiction isn’t theoretical. Colorado runs OmniSalud, a health insurance subsidy program exclusively for undocumented immigrants. In November 2025, the state used the “OmniSalud Lottery”—a random selection process—to determine which 6,700 of 12,000 enrolled undocumented immigrants would retain subsidized healthcare for 2026. The lottery makes consequential decisions about healthcare access.
It processes applications through Colorado Connect, an automated online platform. And it explicitly discriminates based on immigration status—you can only qualify if you’re undocumented. This is algorithmic discrimination based on national origin in healthcare decisions. The exact behavior Colorado’s AI Act prohibits, and the Attorney General is empowered to fine at $20,000 per violation.
The state created a law requiring companies to prove their AI doesn’t discriminate by national origin, then deployed an automated system that discriminates based on national origin. Colorado isn’t just requiring discrimination to prevent discrimination. It’s doing the discrimination.
Two Incompatible Visions
The split reveals competing philosophies about AI’s legal status.
The personhood approach—Missouri, Idaho, Utah, Washington—establishes clear legal boundaries using existing frameworks. AI isn’t a person. Liability stays with humans. These laws are short, simple, preventative. They clarify what AI cannot be before capabilities advance.
The bias-monitoring approach—Colorado, California, Illinois—imposes new demographic tracking requirements and creates new compliance regimes. These laws are long, complex, reactive. They mandate what AI must do based on outcomes measured by protected characteristics.
Congress remains paralyzed. Over 150 AI bills introduced since 2023, none became law. The House attempted a 10-year moratorium on state AI regulation in the 2025 budget bill. The Senate stripped it by a 99-1 vote.
President Trump rescinded President Biden’s AI executive order. Federal agencies offer guidance but no binding rules. The vacuum forced states to act.
The EU’s AI Act, effective March 2026, originally considered granting AI “electronic personhood” but rejected it over liability concerns. Europe chose to regulate AI as a tool, keeping responsibility with humans—closer to Missouri’s philosophy than Colorado’s.
State attorneys general now rely on existing anti-discrimination laws. But guidance isn’t law. And laws that contradict each other aren’t a framework.
What This Reveals
These contradictory laws expose American policymaking’s fundamental confusion about artificial intelligence. We cannot decide whether AI is dangerous because it might gain rights, or dangerous because it might violate ours.
Missouri keeps it simple. AI isn’t a person. The Declaration’s language about being “endowed by their Creator with certain unalienable Rights” doesn’t apply, even though AI has numerous human creators. Rights belong to humans. Responsibilities belong to humans.
Colorado demands companies engage in the very discrimination they’re trying to prevent. It creates compliance regimes requiring demographic tracking while claiming to oppose demographic-based decisions. It makes color-blind AI systems illegal.
A company cannot simultaneously comply with laws saying “never consider race” and laws saying “always monitor racial outcomes.” This isn’t a technical problem. It’s a philosophical contradiction.
As more states adopt one model or the other, businesses operating nationally face impossible choices. Financial institutions must decide whether to track borrower race—Colorado requires it, fair lending laws traditionally forbid it. Employers must choose whether to monitor hiring outcomes by protected class—California mandates it, Title VII traditionally prohibits it.
The personhood question and the discrimination question are both unsettled. But states are writing answers anyway—contradictory answers that reveal policymakers don’t actually know what problem they’re solving.
Missouri’s approach prevents AI from gaining the rights of persons. Colorado’s approach prevents AI from violating the rights of persons. Both cannot be true simultaneously unless we fundamentally redefine what “rights” means in American law.
Until Congress acts, American AI policy remains what it’s always been: multiple creators, no clear rights, no coherent responsibility. The Missouri Compromise and the Colorado Paradox aren’t just state experiments. They’re evidence of a larger failure—the inability to answer basic questions about technology’s role in a republic built on natural law and individual rights.
The Declaration says all men are created equal and endowed by their Creator with unalienable rights. AI has many creators. But Missouri says that doesn’t matter. Colorado says it does—just not in the way anyone expected.
Support Independent Local Journalism — High Country Advocate was created as a real alternative to regional media that too often silences dissenting voices while taking sides in the political struggle. Producing in-depth, unflinching reporting like this series and others is expensive: servers, editing, research time, and legal review all add up quickly. If these articles have informed you or given you new perspective, please consider supporting HCA with a paid subscription — every subscriber helps keep this reporting strong and independent. 