US State AI Laws Compared
Four states and cities have enacted substantive AI compliance laws. Here's how Colorado, New York City, Illinois, and California compare on scope, obligations, and enforcement.
Side-by-Side Comparison
| Attribute | CO SB 24-205 | NYC LL 144 | IL AIVIRA | CA AB 2013 |
|---|---|---|---|---|
| Jurisdiction | Colorado (statewide) | New York City | Illinois (statewide) | California (statewide) |
| In effect since | June 30, 2026 | July 5, 2023 | January 1, 2020 | January 1, 2026 |
| Primary focus | High-risk AI in 8 consequential decision domains | AI hiring/promotion tools (AEDTs) | AI video interview analysis | Generative AI training data transparency |
| Who must comply | Deployers + developers of high-risk AI affecting CO consumers | Employers/agencies using AEDTs for NYC-based candidates | Employers using AI video analysis for IL applicants | Generative AI developers above compute threshold |
| Key obligation | Impact assessment + consumer appeal right | Annual bias audit + public disclosure | Consent + notice + deletion rights | Training data transparency disclosure |
| Max penalty | $20,000/violation | $1,500/day/violation | Actual damages | Actual damages |
| Private lawsuit? | No | No | Yes | Yes |
| Enforcement agency | Colorado AG | NYC DCWP | Courts only | California AG + Courts |
| Sector scope | All sectors using high-risk AI | Employment only | Employment (video) only | AI developers (generative) |
Who Needs to Worry About What
HR Tech / Hiring Platforms
If you offer automated resume screening, candidate scoring, or video interview analysis to employers, you need annual bias audits (NYC) and consent/deletion workflows (Illinois). These apply to your customers' use of your platform.
Generative AI Developers / LLM Companies
If you train large language models or image generation models and make them available to US or EU users, California AB 2013 requires training data transparency disclosure. EU AI Act adds additional GPAI requirements.
Enterprise AI Deployers (Any Sector)
If you deploy AI that makes consequential decisions affecting Colorado consumers in employment, credit, healthcare, or housing, the Colorado AI Act requires impact assessments and consumer disclosures from June 2026.
Financial Services
Financial institutions deploying AI for credit, insurance underwriting, or employment screening face the most overlapping obligations — Colorado's high-risk AI rules, NYC's hiring AI audit requirements, and EU AI Act high-risk classification.
Key Insight: The Patchwork Problem
Unlike the EU AI Act — which creates a unified framework across 27 countries — US AI compliance is a growing patchwork of state and local laws, each with different triggers, obligations, and enforcement mechanisms. A single AI product used across the US may be subject to NYC's bias audit requirements, Colorado's impact assessment mandate, Illinois's consent and deletion rules, and California's training data disclosure requirements — simultaneously.
The practical solution is to build the most demanding requirement into your baseline. A compliance program designed for the Colorado AI Act and NYC LL 144 simultaneously — combined with California AB 2013 disclosure documentation — will satisfy most US AI compliance obligations as additional states pass laws.