January 24, 2025 - 7:00pm

Republicans are in a victorious mood. In his first week in office, Donald Trump has issued an executive order banning diversity, equity and inclusion (DEI) initiatives from the federal government, cancelled grants to censor “disinformation”, and repealed Joe Biden’s executive order on AI safety regulation.

But while things look promising at the federal level, there are a series of state bills in progress which could force AI to subscribe to DEI policies. The most important of these is happening in Texas. Many top tech companies have moved from Silicon Valley to the Lone Star State in pursuit of freedom and less regulation, but this could be undermined by the Texas Responsible AI Governance Act (TRAIGA). “As we celebrate in DC, Texas is about to hand its AI industry to Biden bureaucrats and Big Tech,” said Palantir co-founder Joe Lonsdale, adding that TRAIGA “would kill innovation, jobs, and let China win”.

It is not only innovation, however, that will suffer; this could also usher in a new age of censorship. The Biden administration’s legislation promoted “equity in AI” and has funded groups which are now pushing to enact “disparate impact” laws. These laws are designed to safeguard against “unintended discrimination” when AI makes decisions about or relating to employees.

Disparate impact is the idea that legally protected categories of people can suffer more than others from rules or legislation that, on the surface, appear neutral. Essentially, the same rules leading to different outcomes is an example of disparate impact which, in most cases, people assume is evidence of racism. But AI, in thrall to progressive nostrums, has ended up in ridiculous places by trying to get rid of possible disparate impacts. So much so that Google Gemini last year generated misleading images of black Nazis and Native American female Founding Fathers. Its AI model, which was specifically trained to comply with Biden’s executive order, racially balanced all images, so as to avoid different outcomes between protected groups.

What is strange about this Texas AI bill, which would undermine Trump’s moves at the federal level, is that it was tabled by Republican State Representative Giovanni Capriglione. It is less strange, though, when it is known that the Future of Privacy Forum (FPF), a think tank that received several government grants during the Biden administration, hosts a working group coordinating these bills with Capriglione on its steering committee. It received $1 million from the State Department, $500,000 from the National Science Foundation, and $200,000 from the Department of Energy. FPF operates a working group dominated by Democrats, trying to pass the same bill in Republican-led states. Until Thursday, its website read that it “intended to advance the National Strategy and [Biden] EO on AI” and “promote equity”. The website removed these phrases later that day.

TRAIGA is one of over 10 state bills introduced by members of FPF’s working group, though FPF’s CEO now denies drafting or supporting it. These bills, which are almost identical, would put the burden on all AI users — which they call “deployers” — to avoid disparate impact “harms” in large parts of the economy. Those working in education, hiring, finance, government, healthcare, law, and numerous other industries would be regulated by these proposed laws. Some versions of the bill, including the Texas one, would also create a new regulatory agency to enforce these laws. It would enable bureaucrats to set arbitrary guidelines for what users must comply with. In either case, both these bills would create a new complex of AI censorship and compliance, similar to the social media censorship Trump is now working to reduce.

Republicans need to remember that the Biden administration spent four years building up powerful networks across the country to continue pushing DEI and censorship. While Republicans may have won the White House, the House of Representatives and the Senate, Biden’s machinery continues to operate in many state Houses.


Brian Chau is a mathematician, software engineer, and independent writer at cactus.substack.com.

psychosort