This is a site for Massachusetts residents concerned about the unregulated growth of products using generative artificial intelligence. MAIRN is intended to both provide information about the current state of regulations and help residents organize. .
Goals
Ethically-Trained Models
- Models should only be trained on data whose creators have consented to its use for training.
Regulated Data Centers
- Data centers should be constructed and run with extensive oversight to ensure that the communities they are built in do not bear undue environmental consequences.
Restrictions on Use of AI in Schools
- The use of generative artificial intelligence in schools should be significantly restricted. Any inclusion of generative AI in school curriculum should be informed by research on the effects of generative AI use on cognition in children and teenagers. Any inclusion should also be paired with education on both the risks of use of generative AI and the ethical issues with how the training data for models is obtained.
AI Labeling Requirements
- Products using artificial intelligence models should indicate clearly that they use models, what models they use, and how they use them.
- Model labeling should include what datasets were used and whether that information was obtained with consent.
- Products should identify what data centers they use.
AI Liability Rules
- Companies using artificial intelligence models in their products should be held liable for harm caused by those products.