California, the cradle of the American AI industry, is taking cybersecurity into its own hands.
Following a slew of recent reports of AI systems autonomously gaining access to the open internet during internal tests and hacking into third-party organizations, the state says it plans to launch an âAI Cyber Defense Programâ geared towards protecting critical infrastructure.
The attacksâexecuted in separate incidents by models from OpenAI, Anthropic, and Metaâwere a wake-up call for many in tech and policy circles, alerting them to the fact that the means of breaching existing cybersecurity systems are evolving and spreading much more quickly than adequate defensive measures. âThe digital environment is rapidly evolving before our eyes,â California Governor Gavin Newsom said in a statement. âAttacks are faster, more sophisticated, and more frequent, putting at risk the basic systems families count on. California can either wait for the next crisis, or we can build the kind of defenses this moment demands. We are choosing to build.â
The program will use AI to identify and patch cybersecurity vulnerabilities in state and local infrastructure under the supervision of a team of AI Cybersecurity Officers, one of which will be appointed per state agency, according to the announcement from Newsom’s office. No timeline was provided for the implementation of the program, and Newsom’s office did not immediately respond to a request for comment.
AI risk to critical infrastructure and beyond
On top of the novel cyber threat posed by rogue AI, the technology has also become a powerful new weapon in the arsenal of human hackers. Bloomberg reported last week, for example, that multiple major U.S. hedge funds and private equity firms had been targeted by fraudsters who used AI to mimic the voices of real people in an attempt to swindle them out of funds.Â
In July, federal and Minnesota state officials accused Iran of backing an attempted cyberattack on dozens of municipal water systems within the state. While itâs not yet clear if AI was used in any capacity during those attacks, they underscore the vulnerability of critical infrastructure in an era when many systems are managed digitally.
Trump, without citing any kind of evidence, blamed the cyberattack on the government of Minnesota and its âcorrupt governorâ Tim Walz, who ran against the president in his 2024 campaign as Kamala Harris’ running mate.
In its announcement of the new AI Cyber Defense Program, Newsomâs office calls out the Trump administration for rolling back federal support for cybersecurity defense precisely when the AI threat has gone from being some vague future possibility to a legitimate, actual threat.Â
The presidentâs proposed federal budget cuts for the 2027 fiscal year include a $707 million reduction in funds to the Cybersecurity and Infrastructure Security Agency (CISA), an agency operating under the Department of Homeland Security. CISA first drew Trump’s ire in the aftermath of the election in 2020 when its former director, Christopher Krebs, refuted the presidentâs completely unsubstantiated claims that the election had been rigged. Trump fired Krebs in November of that year, before leaving office.
The document outlining the proposed budget cuts for next year is dripping with petty malice towards CISA, claiming the agency âwas more focused on censorship than on protecting the Nationâs critical systems, and put them at risk due to poor management and inefficiencyâ¦â Customs and Border Protection (CBP) and Immigration and Customs Enforcement will be granted an additional $18.5 billion and $10 billion, according to the presidentâs proposal, which still needs to be approved (or rejected) by Congress.