Colorado: The First State to Sign an AI Regulation Bill

On May 17, 2024, Colorado took a groundbreaking step by enacting SB24-205, also known as the Consumer Protections for Artificial Intelligence Act (Colorado AI Act). Governor Jared Polis's signature on this bill makes Colorado the first state in the United States to adopt comprehensive legislation targeting artificial intelligence (AI).

Overview of the Colorado AI Act

The Colorado AI Act addresses AI's increasing influence in various aspects of life and ensures consumer protection. This legislation establishes stringent requirements for the developers and deployers of "high-risk artificial intelligence systems." These AI systems significantly impact decisions related to essential services such as education, employment, financial services, healthcare, housing, insurance, and legal services.

Key Definitions

• High-Risk AI Systems: AI systems that make or significantly influence consequential decisions affecting consumers' access to or the cost and terms of vital services.
• Algorithmic Discrimination: Unlawful differential treatment resulting from the use of AI, disadvantaging individuals based on protected characteristics such as age, race, disability, etc.

Obligations for Developers and Deployers

The law imposes several obligations on AI developers and deployers to safeguard against algorithmic discrimination and ensure transparency:

For Developers

1. Reasonable Care: Developers must use reasonable care to protect consumers from known or foreseeable risks of algorithmic discrimination.
2. Disclosures and Documentation: Developers must provide detailed documentation about the data used, system limitations, intended use, and measures to mitigate discrimination risks.
3. Public Statements: Developers are required to publish information about their high-risk AI systems and how they manage discrimination risks on their websites.
4. Reporting to AG: Developers must report any known discrimination risks to the Attorney General and all deployers within 90 days of discovery.

For Deployers

1. Risk Management Policy: Deployers must implement a comprehensive risk management policy that aligns with the NIST AI Risk Management Framework or other recognized standards.
2. Impact Assessments: Deployers must conduct impact assessments for high-risk AI systems at least annually and after any substantial modifications.
3. Consumer Notices: Deployers must notify consumers when high-risk AI systems are used in making consequential decisions, providing detailed information about the system and the decision-making process.
4. Disclosure of Discrimination: If a high-risk AI system causes algorithmic discrimination, deployers must report this to the Attorney General within 90 days.

What This Means for Implementing an AI Chatbot on Your Website

The Colorado AI Act has significant implications if you're considering implementing an AI chatbot on your website. Here’s what you need to consider: Your chatbot must inform users that they are interacting with an AI system unless it’s obvious, which builds trust and ensures compliance with transparency requirements. You need a robust risk management policy to address potential algorithmic discrimination and other risks, involving regular reviews and updates to your AI systems. Regular assessments of your chatbot’s impact are essential, especially if your AI system influences consequential decisions, helping identify and mitigate biases or unfair treatment. Additionally, ensure that your users have the ability to understand and challenge decisions made by your AI chatbot, including correcting any incorrect data and appealing adverse decisions.

How await.ai Can Help

At await.ai, we specialize in helping highly regulated organizations deploy ethical, safe, and unbiased AI. Our AI chatbot solution, Await Cortex, comes with comprehensive safeguards out-of-the-box to ensure compliance with regulations like the Colorado AI Act. Here’s how we can assist you:

• Compliance Safeguards: We implement strict safeguards to protect against algorithmic discrimination and ensure transparency in AI interactions.
• Human Approval Workflows: Await Cortex requires human oversight to review and approve AI decisions, ensuring accuracy and fairness.
• Access to Training Data: We provide detailed documentation on the data used to train our AI systems, helping you understand and manage any potential biases.
• Adaptive Disclaimers: Await Cortex allows you to create automatic disclaimers in your chatbot based on keyword flags to protect yourself.

If you want to learn more about ensuring your AI systems are legally compliant, contact await.ai for a free consultation and a demo of our product, Await Cortex, designed to protect your organization from legal risks.