In Sacramento, California, efforts to establish groundbreaking safety measures for the largest artificial intelligence systems have taken a significant step forward. The proposed legislation, designed to minimize potential risks associated with AI, would mandate companies to test their models and publicly disclose their safety protocols. This is crucial to prevent the manipulation of AI models that could lead to catastrophic events like the shutdown of the state’s electric grid or the development of chemical weapons.

Despite facing opposition from various tech giants and venture capital firms, the bill managed to secure a narrow victory in the Assembly. If it passes the final Senate vote, it will then go to Governor Gavin Newsom for approval.

The proposed legislation, authored by Democratic Sen. Scott Wiener, is seen as a crucial step in establishing safety regulations for large-scale AI models in the United States. It specifically targets systems that require more than $100 million in data to train, a threshold that no current AI models have reached.