AI just wrote a bill to regulate itself

From: POLITICO's Digital Future Daily - Wednesday Jul 19,2023 08:39 pm
How the next wave of technology is upending the global economy and its power structures
Jul 19, 2023 View in browser
 
POLITICO's Digital Future Daily newsletter logo

By Mohar Chatterjee

With help from Derek Robertson

Massachusetts State House

The Massachusetts State House | Lisa Kashinsky/POLITICO

The question keeps coming up: When will robots write our laws?

What will happen when they do?

We’re about to find out.

Back in January — a week before Rep. Jake Auchincloss (D-Mass.) read a ChatGPT-generated speech on the House floor, marking the first high-profile use of the technology in Congress — an enterprising state senator in Massachusetts asked the powerful AI chatbot to go a step further and actually write part of a law to regulate its own use, as POLITICO's Kelly Garrity and Lisa Kashinsky reported last week in Massachusetts Playbook.

His office introduced it soon afterward, and last week Massachusetts lawmakers on Beacon Hill heard testimony on the bill, called “An Act drafted with the help of ChatGPT to regulate generative artificial intelligence models like ChatGPT.

DFD spoke with state Sen. Barry Finegold, the politician behind the bill — and one of its two official human co-sponsors — about how it went, and what the process means for lawmaking, democracy and, um, humans.

(View a screenshot of the ChatGPT prompt and ensuing bill here.)

Coaxing ChatGPT to write a law wasn’t all smooth sailing, an aide from his office told me. Getting the bill up to snuff required multiple prompts and language cleanup, since the AI was not immediately familiar with the format of Massachusetts General Laws.

But the exercise had a purpose besides just the bill itself.

“AI is going to be part of every single thing you do,” Finegold said — much like the internet. The point of using the ChatGPT to write a law, he said, was to focus legislative attention on the emerging technology faster than state or federal legislatures usually move.

Finegold sees social media as a cautionary tale in which lawmakers didn’t move nearly fast enough to keep its effects in check. He doesn’t want to make the same mistake with AI. “I fault us, the government, a lot of times for not putting up the basic guardrails and the rules of engagement that we should have had in place years ago,” he said.

The AI-written bill, modeled on a longer data privacy law also introduced by Finegold, hits some of the core points in the AI debate. It addresses the problem of plagiarism, requiring companies to implement a watermarking system on AI-generated content. AI companies are also required to disclose their algorithms and data collection practices to the state attorney general’s office, as well as run regular risk assessments on AI systems.

He had some words for Washington’s role in all this, or lack of a role. “I do think the federal government should be doing this — that we in the state level should not be,” Finegold said. The senator is looking for “a real bipartisan push in Washington, because unlike privacy you really can have fifty different AI policies in fifty different states. It’s just not gonna be productive.”

But the AI bill may hit the same stumbling blocks in Beacon Hill that emerging tech laws face in any legislature — including Congress.

Is this law going to pass? POLITICO's Kashinsky told me the state “doesn’t seem to be in any rush to regulate artificial intelligence.” She expects perhaps a commission to study it, an idea in another bill.

Finegold says he’s “hopeful” that his data privacy legislation will be passed this term. He is focused on pushing that bill past the finish line of the state legislature first. “Data privacy is more of a priority just because, unfortunately, we've seen the federal government is unable to really get a privacy bill done,” the senator said. “Other states have passed privacy bills and we feel like we need something here in Massachusetts.”

On AI, what he wants at a federal level is for Congress to “pass a similar bill that we proposed,” Finegold said. The set of policy points in the Massachusetts AI bill are ones that businesses “can clearly put their arms around,” he said.

This time around, the tech sector itself is asking for rules, the senator pointed out, referencing OpenAI CEO’s Sam Altman’s public messaging on needing AI guardrails. It’s up to the federal government now to draft an AI law that people can “live with in the business sector, but at the same time will protect consumers,” he said.

 

JOIN 7/11 FOR A TALK ON THE FAA’S FUTURE: Congress is making moves to pass the FAA Reauthorization Act, laying the groundwork for the FAA’s long-term agenda to modernize the aviation sector to meet the challenges of today and innovate for tomorrow. Join POLITICO on July 11 to discuss what will make it into the final reauthorization bill and examine how reauthorization will reshape FAA’s priorities and authorities. REGISTER HERE.

 
 
a top spy on ai

LONDON, ENGLAND - MAY 17: A rainbow flag flies alongside a Union Jack outside the MI6 building in support of International Day Against Homophobia on May 17, 2016 in London, England. The rainbow flag is commonly associated with gay rights and the International Day Against Homophobia started in 2004, with people across the globe celebrating gay rights. The MI6 building at Vauxhall is the headquaters of the British Secret Intelligence Service (SIS). (Photo by Chris Ratcliffe/Getty Images)

The MI6 building. | Getty Images

The U.K.’s top intelligence officer sees an immediate, practical use for AI in intel-gathering and national defense.

And also an immediate, practical threat from the other side.

Speaking today with POLITICO’s Anne McElvoy, MI6 chief Richard Moore said that as much as AI might be a helpful tool for data collection or analysis, it’s also paramount for his agency to anticipate and counter how its adversaries might use it.

“It will be a significant part of our role going forward into the future to try and uncover, detect, and disrupt people who would like to develop AI in directions which are dangerous,” Moore said. “It’s really important that we work to preserve human agency over the technologies we’re developing, and not all actors out there may approach this with the same degree of responsibility that we in the U.K. do.” — Derek Robertson

a senator takes on ai

Gary Peters speaks during a hearing on Capitol Hill.

Sen. Gary Peters (D-Mich.). | Francis Chung/POLITICO

One Democrat in the Senate is hoping to make sense of all the legislative theater happening on Capitol Hill around AI.

A report from POLITICO’s Brendan Bordelon today unpacks the steady maneuvering Sen. Gary Peters (D-Mich.) has put into legislation about the technology — including two bills that have reached the Senate floor this year and another getting marked up today. Peters’ approach hasn’t led to sweeping, European-style proposals for guardrails around what AI can or can’t do. Instead, he’s focused on tweaking around the edges to better prepare the federal machinery for what seems inevitable: introducing bills that require greater transparency around the government’s use of AI, for example, and requiring training for government employees involved in AI procurement.

Talking about his transparency bill, Peters told Brendan: “It allows us to test some of these ideas and see how it actually works in practice, and that could be a model for what we want to do broadly in the commercial side.” The newest bill, the AI LEAD Act, would require federal agencies to appoint a “chief AI officer” and create a council of those officers to meet regularly and develop an overall agency-level approach to the technology. — Derek Robertson

D.C. RETHINKS REENGINEERING THE PLANET

There have been plenty of sci-fi solutions to climate change proposed over the years, from pumping aerosols into the atmosphere to partially blocking out the sun itself.

After a cautiously positive report about geoengineering from the White House earlier this year, however, lawmakers are decidedly cooling on the idea. (Sorry.) As E&E News’ Emma Dumain and Corbin Hiar reported yesterday, there’s little appetite among congressional Democrats to seek out future-tech fixes to climate change when America has hardly begun to try the more conventional ones.

“Why try a complicated and potentially dangerous solution when we have the obvious one right in front of us?” asked Sen. Sheldon Whitehouse (D-R.I.), referring to efforts to sharply cut carbon emissions in the first place.

Rep. Sean Casten (D-Ill.) used the enthusiasm for geoengineering among some climate change activists as a motivation to get to reducing emissions, saying “At some point, someone is going to do this, and if we haven’t done the research to determine what are the consequences of doing this are — what are the most responsible ways to do this, how do you think about the trade-offs — then we can’t really complain if someone else does this research and decides to do something.” — Derek Robertson

Tweet of the Day

A new LLM truthfulness benchmark just dropped. (Context: Alabama in fact has a higher per capita GDP than Japan.)

the future in 5 links

Stay in touch with the whole team: Ben Schreckinger (bschreckinger@politico.com); Derek Robertson (drobertson@politico.com); Mohar Chatterjee (mchatterjee@politico.com); and Steve Heuser (sheuser@politico.com). Follow us @DigitalFuture on Twitter.

If you’ve had this newsletter forwarded to you, you can sign up and read our mission statement at the links provided.

 

SUBSCRIBE TO POWER SWITCH: The energy landscape is profoundly transforming. Power Switch is a daily newsletter that unlocks the most important stories driving the energy sector and the political forces shaping critical decisions about your energy future, from production to storage, distribution to consumption. Don’t miss out on Power Switch, your guide to the politics of energy transformation in America and around the world. SUBSCRIBE TODAY.

 
 
 

Follow us on Twitter

Ben Schreckinger @SchreckReports

Derek Robertson @afternoondelete

Steve Heuser @sfheuser

 

Follow us

Follow us on Facebook Follow us on Twitter Follow us on Instagram Listen on Apple Podcast
 

To change your alert settings, please log in at https://www.politico.com/_login?base=https%3A%2F%2Fwww.politico.com/settings

This email was sent to by: POLITICO, LLC 1000 Wilson Blvd. Arlington, VA, 22209, USA

Please click here and follow the steps to .

More emails from POLITICO's Digital Future Daily

Jul 18,2023 08:03 pm - Tuesday

The religious mystery of AI

Jul 17,2023 08:09 pm - Monday

Musk's big bet on anti-'woke' AI

Jul 14,2023 08:01 pm - Friday

5 questions for Paul Barrett

Jul 13,2023 08:02 pm - Thursday

Toward humanity's off-planet future

Jul 12,2023 08:08 pm - Wednesday

Algorithms get a new watchdog

Jul 11,2023 08:18 pm - Tuesday

Europe’s agenda for… not 'the metaverse'

Jul 10,2023 08:01 pm - Monday

Government-issued digital money gets closer