to Wiener citing “significant concerns” about the Safe and Secure Innovation for Frontier Artificial Intelligence Models Act. In the five-page document, Rob Sherman, the vice president and deputy chief privacy officer of policy for Meta, slammed the proposal for “creating a mandate to meet standards that don’t exist” and placing “unrealistic obligations on open source providers.”
Meta shares Wiener’s goals to advance “the safe and responsible development of the most capable AI models,” Sherman noted, but the state senator’s execution of the objective misses the point.“There are no existing benchmarks for the types of AI ‘safety testing’ that the bill contemplates,” Meta wrote. “Industry and governments are working together to determine which benchmarks and frameworks for testing and evaluating models can and should be used.
Meta also said the bill would thwart small businesses’ capability to harness the technology, stating it “discourages a key avenue for small businesses and startups to compete and participate in the innovation ecosystem ⏤ the very environment California has led the world in fostering.” The comments are a marked departure from the company’s historic posture, representing the first time Meta has publicly undermined AI regulation at the state level.on Tuesday, pushing back on the idea that his legislation would harm small businesses. “It doesn’t cover startups. It doesn’t ban open sourcing,” the San Francisco senator responded to Meta.Wiener added he is “grateful” for feedback and will continue to “engage with diverse stakeholders” while ending his statement with a warning.
“We also need to have eyes wide open that some in the tech sector are opposed to any and all forms of AI regulation, no matter how limited or narrow,” Wiener said.