On Sept. 29, Gov. Gavin Newsom signed into law Senate Bill 53, the Transparency in Frontier Artificial Intelligence Act (TFAIA). Enhanced whistleblower protection for key employees of companies, called large frontier developers, that train AI models on an exceptional level of computing power and that have over $500 million in annual revenue, is one of several guardrails the new law installs against the catastrophic risks of using and deploying powerful AI systems.
A risk is “catastrophic” under TFAIA if it is foreseeable that development, storage, use, or deployment of a frontier model will materially contribute to the death of, or serious injury to, more than 50 people or cause more than $1 billion in property loss or damage from a single incident in which the model: (1) provides expert-level assistance in the creation or release of chemical, biological, radiological or nuclear weapons; (2) engages in conduct without meaningful human oversight that is either a cyberattack or that would constitute murder, assault, extortion or theft if committed by a human; or (3) evades the control of the model’s developer or user.
The effective deployment of sophisticated AI systems may bring great rewards, but also may carry great risks. “‘Foundation’ or ‘frontier’ models are the largest, most powerful AI systems being built today,” wrote a Senate Rules Committee analyst. “A frontier model might be used to cure disease or, conversely,...
When examiners don’t get veterans’ records, bad decisions aren’t a mistake—they’re a systemic failure. In a blistering oversight hearing this week, members of Congress openly questioned whether th...