Artificial Intelligence Regulation - Three Months of… What?

A lot of small things have happened in this area since mid-May.  To what do they add up?

First, “Trump Cancels Signing of A.I. Executive Order” (Tripp Mickle and Sheera Frenkel, The New York Times, May 21^st).  Another flip-flop for our president, the kind of thing that happens when he finds out only after proposing something that it won’t work.  It was “an executive order that would give the government the power to evaluate artificial intelligence models before they were publicly released.”  The cancellation was the right decision, as nothing and no one in our government could possibly do that competently.

Next, “Elizabeth Warren calls for taxing AI industry to ‘invest in people’” (Alex Nitzberg, Fox Business, about May 28^th).  As I wrote in Work’s New Age 15 years ago, we may need to rearrange the sources of our tax revenues as they change in significance - for example, if the number of people working becomes greatly reduced, from efficiency, foreign competition, and automation, it would become unfair to count on them for the same share of tax collections they currently provide.  If, as Warren also said, “the tax system incentivizes replacing workers with AI,” that needs correction as soon as possible.  It is too soon for directly taxing AI to be noncontroversial, but, if AI shows clearer signs of replacing large numbers of workers, we will need to discuss that with rather more urgency.  Accordingly, Warren is right to bring up this idea now.

Moving along, per a “Fox News Poll:  Voters see AI regulation as urgent, rank safeguards ahead of innovation” (Victoria Balara, Fox News, May 28^th).  Innovation is great, but there are always guardrails, and, as we should have learned from dealing with the likes of Uber and Airbnb, using new technology should not exempt firms from governmental control.  Here, “nearly 8 in 10 think is it extremely (40%) or very (37%) urgent for the government to address AI regulations.”  There is no real partisan divide there, as 83% of Democrats, 77% of Republicans, and 82% of Independents want “to prioritize protecting the public interest.”  We’ll see how the other pieces in this post support this need.

Less than two weeks after the first article here came out, we saw as “Trump signs AI order that asks companies to give government early access” (Joey Garrison, USA Today, June 2^nd).  In it, the president “asks AI developers to voluntarily submit their models to the federal government to review for potential security risks.”  Although Washington “would have access to advanced AI models submitted for testing for up to 30 days - a shorter period than the 90-day window established in the draft order that Trump shelved,” that’s still a long time in the AI industry, and it is even less clear what our government can constructively accomplish.

Another effort from a different direction appeared as a “New Super PAC Aims to Rally Tech Workers to Help Limit A.I.” (Mike Isaac and Theodore Schleifer, The New York Times, June 18^th).  “Two Democratic operatives are aiming to leverage the unease within the tech industry over A.I. and harness an agitated work force into a political movement.”  We do have that.  The new PAC is smaller than its opposing one, Leading the Future, but “has started deploying” its funds, “buying ads for the Democratic party in New York City’s 12^th Congressional District to support Alex Bores, a former tech worker who has written A.I. safety legislation.”  No results here yet, but this organization will get support from Republicans as well.

Soon afterwards, “U.S. Presses Meta to Agree to A.I. Reviews as Security Concerns Rise
 (Tripp Mickle, Eli Tan and Sheera Frankel, The New York Times, June 23^rd).  Voluntary again, and “the latest example of the administration’s efforts to step up oversight of the A.I. industry after promoting a hands-off approach to” it.  Yet “Meta is the only major U.S. developer of A.I. technology that has not reached an agreement to voluntarily share its models with the federal government for review,” so also for another reason nothing is guaranteed.

Days later, “U.S. Loosens Restrictions on Anthropic’s Mythos A.I. Model” (Sheera Frenkel, The New York Times, June 26^th) was updated into “U.S. Lifts Restrictions on Anthropic’s Most Powerful A.I. Models” (Sheera Frankel and Ana Swanson, The New York Times, June 30^th).  The first “move de-escalates a clash between the Trump administration and the company,” and the second extended that to all the company’s strongest products.  It should surprise no one to read that “Trump officials are still working on a framework for how companies should formally submit new A.I. models for review, and what standards they would be held to.”

On a side not as opposite as it may seem, “AI could unleash ‘single greatest productivity revolution’ if Washington avoids overreach:  report” (Sophia Compton, Fox Business, June 28^th).  AI has great potential, we all can agree, but we don’t see eye-to-eye on what “overreach” would be - especially when it seems the potential overreachers are neither consistent nor effective.

Finally, something substantive!  “New York makes history with first-of-its-kind law regulating AI-powered commercials” (Julie Bonavita, Fox News, July 2^nd).  “The state’s synthetic performer disclosure law, signed by Gov. Kathy Hochul in December 2025, requires advertisements featuring an AI-generated person to include a clear label indicating the individual is not real.”

 That’s it.  I found nothing else.  Now, can we create boundaries for more things, and beyond just one state?  We are collectively floundering at that.  As the survey above shows, we need, somehow, to do better with artificial intelligence regulation - and, given recent intensified concerns, do it soon.  That is our bipartisan task.