A second California state senator has launched payments meant to manage AI techniques, notably these utilized by state companies.
Senator Steve Padilla, a Democrat, introduced Senate Bills 892 and 893, establishing a public AI useful resource and making a “protected and moral framework” round AI for the state. Senate Invoice 892 would require California’s Division of Expertise to develop security, privateness, and non-discrimination requirements round companies utilizing AI. It additionally prohibits the state of California from contracting any AI companies “until the supplier of the companies meets the established requirements.”
The opposite invoice, SB 893, would set up a California AI analysis hub that would offer entry to compute sources and knowledge for teachers.
Padilla’s payments comply with an earlier one introduced by California Senator Scott Wiener in September of final 12 months. That invoice proposed transparency requirements for “frontier” fashions, in addition to safety measures and the institution of a state AI analysis middle. That invoice was categorised as an intent invoice that required additional improvement earlier than it might cross.
A number of federal policymakers have proposed laws regulating AI. The AI Foundation Model Transparency Act would mandate AI firms disclose coaching knowledge sources; the No Fakes Act plans to standardize likeness guidelines to guard people from digital replicas with out their consent; and there are a number of payments that may ban the usage of AI in election marketing campaign adverts. However no main federal laws have handed particularly addressing current issues about AI security, regardless of an government order across the expertise from the Biden administration.
This has led a number of state lawmakers to file state-level legislation. As reported by the Associated Press, Texas, West Virginia, North Dakota, Louisiana, and Puerto Rico have ordered the monitoring or examine of AI techniques utilized by state companies. In the meantime, Connecticut has determined to evaluate all AI companies it makes use of and plans to frequently evaluate these techniques.