← All home signals
Governance

The Explained Denial

A renter turned down by an automated system in Colorado gets a reason within 30 days and a human review. Critics say the same law dropped the duty to test.

What is it?

When software helps turn down a tenancy or insurance application, Colorado law will require notice, a plain-language reason within 30 days and a human second look. The same law dropped the duty to test the system first, so explanation stands where the audit was.

Why now?

Colorado signed SB26-189 on 14 May 2026, replacing its 2024 AI Act before that law took effect in June. Risk assessments, annual reviews and testing duties fell away. Meanwhile, by one civil-rights count, more than 90 percent of landlords screen applicants, and federal disparate-impact enforcement is being withdrawn.

Who’s building it?

Screening-software vendors, landlords, insurers, the Colorado attorney general, civil-rights groups and labour unions contesting the rules.

Colorado legislatureScreening vendorsLandlordsAFL-CIOCivil-rights groups
The full read
Why now, who is building it, where the market goes, and what stays unsettled.
Get weekly →
01 · The market
Who’s in it
Builders — vendors of automated tenant-screening and insurance-decision tools.
Buyers — landlords and insurers who deploy the tools and now owe notices and reviews.
Gatekeepers — the Colorado attorney general, enforcing through consumer protection law.
What’s at stake
Whether a plain-language reason after the fact can stand in for testing a screening model before it is used.
Whether renters can contest a wrongly matched record.
Whether human review changes outcomes or only confirms them.
What’s unsettled
What counts as meaningful human review once the law takes effect.
Whether other states copy the explanation model without the audit.
Whether screening vendors keep testing voluntarily now that Colorado law no longer requires it.
02 · How the demand matureda progression, not a timeline
Novelty
Landlords adopt automated screening that scores applicants from court and background records.
Retention
Screening becomes near universal, with more than 90 percent of landlords requiring it.
Willingness to pay
Colorado lawmakers trade upfront testing duties for explanations and human review owed to each applicant.
Category
Housing decision rights shift from model audits to individual reasons and appeals.
03 · Where the market goesthe forks, not a prediction
→Explanation as the norm: other states copy Colorado’s notice, reason and review model, and pre-deployment testing fades from housing law.
→Human reviews overturn enough screening errors that deployers start testing models on their own.
→Reviews mostly confirm the software, and renters return to lawmakers asking for the audits back.
→Litigation over dismissed evictions and mismatched records restores validation through the courts instead.
The forks turn on whether a reason and a human second look catch bad screening decisions, or only make them easier to accept.
04 · The Right to a Reason deskadjacent market, moving now
The Explainability Demand
Black-box refusals meet a cultural pushback.
Korean data subjects can request explanations of automated decisions.
Chambers and Partners
The Liability Question
Who answers when an automated decision goes wrong.
Colorado’s 2026 rewrite splits AI discrimination liability by relative fault.
Littler
The Always-On Audit
Compliance work moves into software that runs continuously.
A FinCEN proposed rule would weigh banks’ AI use in anti-money-laundering reviews.
U.S. Financial Crimes Enforcement Network
05 · The Wire
01
Colorado’s new law gives a reason within 30 days.
Colorado General Assembly ↗
02
The new law drops the AI Act’s testing duties.
Colorado Public Radio ↗
The panel on this signal
They agreed on
–The law gives a renter notice, a reason and a human review, but no right to a tested model.
–None of these duties apply before January 2027.
–The testing and impact-assessment duties of the 2024 law did not survive.
The tenant-screening dispute specialist disagreed on
–A reason drawn from a wrongly matched court record only explains the error; disputes need the source data corrected, not narrated.
Know more about the method →
✦How this signal was built
01 · Panel
A Delphi panel of 4 AI analysts, each reasoning independently.
→
02 · Debate
They challenge and revise each other’s reads over two rounds.
→
03 · Evidence
Every claim is tied to a cited, linked source on The Wire.
→
04 · Synthesis
A moderator writes it into one read, disagreement and all.