• Deployers are the entities using the AI system to assist in decision- making or offering the AI system to the end user. [...] CONTENT OF AUDITS AND IMPACT ASSESSMENTS • The pre- and post-deployment impact assessments and audits required in this section must include the following information (at minimum): o provenance and quality of training data and inputs, including ensuring the data and testing are sufficient to address the real-world inputs for which the AI system will be used and to evaluate impacts on various demogr. [...] POST-DEPLOYMENT REQUIREMENTS • Deployers should be required to perform audits and impact assessments at least once a year and whenever there is a material change to the context in which the AI system is used, how the AI system operates, or the data that is used as part of ongoing AI system operations. [...] NOTICES AND TRANSPARENCY • Developers and deployers should be required to make public (on their own websites and in a central repository) a plain-language summary of the results of required audits and impact assessments, details about the data on which the AI system was trained, how the risks were weighed against potential benefits and how risks were mitigated, procedures for human review of the A. [...] • Individuals subject to a consequential use of an AI system should have the right to be notified of the outputs, determinations, or decisions produced and to an explanation of how the AI system produced that output, determination, or decision.
Authors
- Pages
- 14
- Published in
- United States of America
Table of Contents
- ABOUT EPIC 2
- AUTHOR 2
- CONTRIBUTIONS BY 2
- ACKNOWLEDGEMENTS 2
- AI Legislation Scorecard. 3
- 1. STRONG DEFINITIONS 5
- Algorithmsartificial intelligenceautomated decision-making 5
- Algorithmic discrimination 5
- Protected classes 5
- Consequential use of an AI system 6
- Bonus if the definition also includes uses of AI that have the 6
- Note Some bills may refer to this concept as high-risk AI or 6
- Developers 6
- Deployers 6
- 2. BASELINE REQUIREMENTS 7
- 3. PROHIBITS ALGORITHMIC DISCRIMINATION 7
- 4. PROHIBITS PARTICULARLY HARMFUL USES 7
- 5. DATA MINIMIZATION 8
- 6. TRANSPARENCY AND ACCOUNTABILITY 8
- A. CONTENT OF AUDITS AND IMPACT ASSESSMENTS 8
- B. PRE-DEPLOYMENT REQUIREMENTS 9
- C. POST-DEPLOYMENT REQUIREMENTS 10
- D. NOTICES AND TRANSPARENCY 11
- 7. DATA SECURITY 12
- 8. PROHIBITS UNFAIR PRACTICES 12
- 9. INDIVIDUAL RIGHTS 13
- 10. ENFORCEMENT AND RULEMAKING 14
- Bonus if the bill allows additional damages for 14
