Cynthia Lummis Proposes Artificial Intelligence Bill, Requiring AI Firms to Disclose Technicals
Senator Cynthia Lummis’s RISE Act of 2025 aims to clarify AI liability for professionals, striking a balance between encouraging innovation and ensuring accountability. The bill focuses on maintaining professional responsibility for advice given, even when AI systems are involved. Crucially, it doesn’t grant blanket immunity to AI developers.
The Act introduces a system of conditional liability protection for AI developers. This protection hinges on the public release of “model cards,” comprehensive documents detailing the AI system’s training data, intended uses, performance metrics, limitations, and potential failures. These cards are designed to aid professionals in evaluating the suitability of the AI tool for their specific tasks. The bill explicitly outlines exceptions to this liability shield. Developers remain liable for recklessness, willful misconduct, fraud, knowing misrepresentation, or actions outside the defined scope of professional use. Furthermore, a duty of ongoing accountability requires developers to update model card documentation within 30 days of deploying new versions or discovering significant flaws, emphasizing continuous transparency.
While the RISE Act stops short of mandating open-source AI models, it allows developers to withhold proprietary information only if it’s unrelated to safety and each omission is justified in writing. This approach attempts to balance the need for transparency with the protection of trade secrets. The debate around open-source AI versus closed-source AI is significant. Concerns have been raised regarding the dangers of centralized, closed-source AI systems, likened to a “god” whose inner workings are opaque and controlled by a few, posing potential risks. The RISE Act seeks to mitigate these risks through detailed disclosure requirements, aiming to promote responsible AI development while preserving professional autonomy in a rapidly evolving technological landscape. The Act’s success will depend on its ability to effectively manage the complexities of AI liability and foster a climate of innovation tempered by accountability.
