California Governor Gavin Newsom issued an executive order earlier in the month to accelerate the development of AI kill switches, safety measures to effectively shut down AI.
This comes amid concerns about AI’s security issues and rapid advancements after Dario Amodei, co-founder of the AI corporation Anthropic, published an open letter on Sept. 12 calling for the slowdown of AI.
The executive order was issued just months after USC Viterbi School of Engineering added an Artificial Intelligence major. According to the school’s website, the major focuses on preparing engineering students for the rapidly evolving AI landscape by designing, training and building models for tasks such as image recognition and language processing.
USC Marshall School of Business also added an AI for Business major in collaboration with Viterbi, with the first cohort entering in August 2023.
“[While] many things may change, this core design principle that guardrails are part of the build and not an after-the-fact observation or consideration, will remain across the board: undergrad, grad, executive, at all levels,” said Kimon Drakopoulos, Associate Professor of Data Sciences and Operations and Vice Dean of AI Initiatives at the USC Marshall School of Business.
In practice, an AI “kill switch” under bill text SB 1047 is controlled by the company that built and runs the model. There would be a set of controls that would halt a training run, pull a model from company servers or revoke the API keys that let other apps reach it, according to the law firm Morgan Lewis.
On the federal level, there have been no actions taken in terms of legislation regarding AI. According to the Brennan Center for Justice, over 150 AI-related bills were presented during the 118th Congress. None of those bills were passed.
Almost all AI-related bills that have been presented in the current 119th Congress are pending, with only two having been passed into law, according to the Brennan Center for Justice. One was a part of the One Big Beautiful Bill Act and it develops AI to accelerate audits for the Department of Defense. The second criminalizes the publication of non-consensual imagery, including images made by AI.
With the inaction from Congress, California has become one of the states with the most AI-related legislation.
“With Donald Trump and Congress asleep at the wheel, Governor Gavin Newsom is once again taking the lead to strengthen AI safety for all Americans,” said Newsom’s statement on his website.
In September 2023, Governor Newsom issued the nation’s first executive order on generative AI, according to an official press release from the Office of the Governor.
Two years later in 2025, he enacted the first major state-safety bill for developers of frontier AI models.
In March of this year, Newsom signed an executive order establishing new trust, safety and vendor certification standards. Earlier this month, he signed further restrictions targeting youth safety and the AI kill switch.
“So I do think that makes sense, but I also don’t think that might be feasible, just because the government is very behind when it comes to their own tech,” said Vanya Shrivastava, a junior studying AI for Business.
Currently, there are no federal laws that require AI companies to report dangerous incidents. California, New York and New Jersey lead the country with the highest number of AI bills introduced.
And President Trump has been supportive of more AI presence. He responded to Dario’s comment by calling concerns over AI a “sick conspiracy” on Truth Social.
“Whether we like it or not, we are living in a world that’s changing at an unprecedented timescale, so your best investment is learning and adapting,” said Drakopoulos.
Edited by Fashion Castillo and Leo Li