Thomas Elias: Trump, Newsom aren’t doing enough to guard against AI threats
It’s high time both President Trump and his frequent sparring partner Gov. Gavin Newsom both realize that the need for better controls on artificial intelligence easily trumps Silicon Valley’s alleged need to create whatever technologies will make the biggest profits.
One of Trump’s final acts of 2025 was signing an executive order for the federal government to take over all regulation of AI, a move spurred in large part by Newsom’s signing a few bills last fall that supposedly put sufficient controls on AI chatbots to keep them from becoming dangerous.
Trump doesn’t want Newsom or anyone else getting involved with this issue, which has seen some forms of AI encourage children toward unhealthy eating and other self-destructive behaviors including suicide. The idea, the president’s aides said, was to prevent states from passing “onerous AI controls” that might allow China to create more advanced forms of this technology than American firms like Google and OpenAI.
Yet, Trump has so far done nothing to rein in any harmful effects of AI.
In reality, his move was just another knee-jerk reaction to prevent Newsom from taking leadership in a significant area and getting another leg up on Republicans in his still undeclared 2028 run for Trump’s current office.
One other reality: Newsom has not done nearly enough to protect Californians from some threats posed by AI, threats anticipated in science fiction stories as long ago as the 1940s.
Yes, he did sign a few new laws that will help a bit here. But there is nothing existential among them.
One new law requires computer and AI suppliers to ask customers for the new user’s age when setting up devices like smartphones and laptops. The suppliers then are supposed to apply the new user’s age to adjust content appropriately.
Another requires some programs to flash warning labels about possible adverse mental health effects of social media posts.
And a third requires chatbots built into many new devices to remind users they are not talking to a human, but rather a machine. The same new law requires suicide prevention personnel to be informed automatically when users show signs of distress in their postings and questions.
But there are no guarantees here that companion chatbots – those that simulate human conversations, like OpenAI’s ChatGPT – will not remain capable of harming humans.
Newsom, under pressure from high-tech companies, vetoed another proposed new law that would have prevented companies from making companion chatbots available to children if they are known to be capable of promoting harmful behaviors like violence, anorexia or other self-harm.
Newsom took severe criticism for vetoing that measure, but argued it was too broad and could prevent children from having any AI access at all.
But even the mild measures he did sign were plainly too much for Trump and some of his corporate supporters.
For one thing, the laws Newsom approved are a tougher package than any other state has adopted, even if they’re still pretty weak. They displease companies like Google, Anthropic, Nvidia and OpenAI, which turned to Trump for an antidote to state regulation in their largest domestic market.
Trump’s response was to order all federal agencies to explore whether they can restrict grants to states that pass any AI regulations at all, of which California is the largest. He did not single out California, rather also criticizing other states. One was Colorado, which last year adopted a law that requires testing of AI programs and then notification of customers if any make consequential suggestions for people in their life decisions.
Rather than caving in to this sort of pressure tactic, Newsom should spend part of this year – his last in state office – encouraging California companies to adopt the kind of protections he vetoed last fall.
If he doesn’t do that, he’ll essentially be bowing to Trump, a situation that’s always made him uncomfortable during the decade or so the two men’s careers have overlapped.
Email Thomas Elias at tdelias@aol.com.