Illinois lawmakers have mixed results in efforts to rein in AI

Chicago Tribune

18 hours ago

Illinois lawmakers have so far achieved mixed results in efforts to regulate the burgeoning technology of artificial intelligence, a task that butts up against moves by the Trump administration to eliminate restrictions on AI.
AI-related bills introduced during the spring legislative session covered areas including education, health care, insurance and elections. Supporters say the measures are intended to address potential threats to public safety or personal privacy and to counter any deceitful actions facilitated by AI, while not hindering innovation.
Although several of those measures failed to come to a vote, the Democratic-controlled General Assembly is only six months into its two-year term and all of the legislation remains in play. But going forward, backers will have to contend with Republican President Donald Trump's administration's plans to approach AI.
Days into Trump's second term in January, his administration rescinded a 2023 executive order from Democratic President Joe Biden, that emphasized the 'highest urgency on governing the development and use of AI safely and responsibly.'
Trump replaced that policy with a declaration that 'revokes certain existing AI policies and directives that act as barriers to American AI innovation.'
Last week, the states got a reprieve from the federal government after a provision aimed at preventing states from regulating AI was removed from the massive, Trump-backed tax breaks bill that he signed into law. Still, Democratic Illinois state Rep. Abdelnasser Rashid, who co-chaired a legislative task force on AI last year, criticized Trump's decision to rescind Biden's AI executive order that Rashid said 'set us on a positive path toward a responsible and ethical development and deployment of AI.'
Republican state Rep. Jeff Keicher of Sycamore agreed on the need to address any potential for AI to jeopardize people's safety. But many GOP legislators have pushed back on Democratic efforts to regulate the technology and expressed concerns such measures could hamper innovation and the ability of companies in the state to remain competitive.
'If we inhibit AI and the development that could possibly come, it's just like we're inhibiting what you can use metal for,' said Keicher, the Republican spokesperson for the House Cybersecurity, Data Analytics, & IT (Information Technology) Committee.
'And what we're going to quickly see is we're going to see the Chinese, we're going to see the Russians, we're going to see other countries come up without restrictions with very innovative ways to use AI,' he said. 'And I'd certainly hate in this advanced technological environment to have the state of Illinois or the United States writ large behind the eight ball.'
Last December, a task force co-led by Rashid and composed of Pritzker administration officials, educators and other lawmakers compiled a report detailing some of the risks presented by AI. It addressed the emergence of generative AI, a subset of the technology that can create text, code and images.
The report issued a number of recommendations including measures to protect workers in various industries from being displaced while at the same time preparing the workforce for AI innovation.
The report built on some of the AI-related measures passed by state lawmakers in 2024, including legislation subsequently signed by Pritzker making it a civil rights violation for employers to use AI if it subjects employees to discrimination, as well as legislation barring the use of AI to create child pornography, making it a felony to be caught with artificially created images.
In addition to those measures, Pritzker signed a bill in 2023 to make anyone civilly liable if they alter images of someone else in a sexually explicit manner through means that include AI.
In the final days of session in late May, lawmakers without opposition passed a measure meant to prevent AI chatbots from posing as mental health providers for patients in need of therapy. The bill also prohibits a person or a business from advertising or offering mental health services unless those services are carried out by licensed professionals.
It limits the use of AI in the work of those professionals, barring them, for example, from using the technology to make 'independent therapeutic decisions.' Anyone found in violation of the measure could have to pay the state as much as $10,000 in fines.
The legislation awaits Pritzker's signature.
State Rep. Bob Morgan, a Deerfield Democrat and the main House sponsor of the bill, said the measure is necessary at a time when there's 'more and more stories of AI inappropriately and in a dangerous way giving therapeutic advice to individuals.'
'We started to learn how AI was not only ill-equipped to respond to these mental health situations but actually providing harmful and dangerous recommendations,' he said.
Another bill sponsored by Morgan, which passed through the House but didn't come to a vote in the Senate, would prevent insurers doing business in Illinois from denying, reducing or terminating coverage solely because of the use of an artificial intelligence system.
State Sen. Laura Fine, the bill's main Senate sponsor, said the bill could be taken up as soon as the fall veto session in October, but noted the Senate has a year and half to pass it before a new legislature is seated.
'This is a new horizon and we just want to make sure that with the use of AI, there's consumer protections because that's of utmost importance,' said Fine, a Democrat from Glenview who is also running for Congress. 'And that's really what we're focusing on in this legislation is how do we properly protect the consumer.'
Measures to address a controversial AI phenomenon known as 'deepfakes,' when video or still images of a face, body or voice are digitally altered to appear as another person, for political purposes have so far failed to gain traction in Illinois.
The deepfake tactic has been used in attempts to influence elections. An audio deepfake of Biden during last year's national elections made it sound like he was telling New Hampshire voters in a robocall not to vote.
According to the task force report, legislation regulating the use of deepfakes in elections has been enacted in some 20 states. During the previous two-year Illinois legislative term, which ended in early January, three bills addressing the issue were introduced but none passed.
Rashid reintroduced one of those bills this spring, to no avail. It would have banned the distribution of deceitful campaign material if the person doing so knew the shared information to be false, and was distributed within 90 days of an election. The bill also would prohibit a person from sharing the material if it was being done 'to harm the reputation or electoral prospects of a candidate' and change the voting behavior of electors by deliberately causing them to believe the misinformation.
Rashid said hurdles to passing the bill include whether to enforce civil and criminal penalties for violators. The measure also needs to be able to withstand First Amendment challenges, which the American Civil Liberties Union of Illinois has cited as a reason for its opposition.
'I don't think anyone in their right mind would say that the First Amendment was intended to allow the public to be deceived by political deep fakes,' Rashid, of Bridgeview, said. 'But … we have to do this in a really surgical way.'
Rashid is also among more than 20 Democratic House sponsors on a bill that would bar state agencies from using any algorithm-based decision-making systems without 'continuous meaningful human review' if those systems could have an impact on someone's civil liberties or their ability to receive public assistance. The bill is meant to protect against algorithmic bias, another threat the task force report sought to address. But the bill went nowhere in the spring.
One AI-related bill backed by Rashid that did pass through the legislature and awaits Pritzker's signature would prohibit a community college from using artificial intelligence as the sole source of instruction for students.
The bill — which passed 93-22 in the House in the final two days of session after passing 46-12 in the Senate on May 21 — would allow community college faculty to use AI to augment course instruction.
Rashid said there were 'technical reasons' for not including four-year colleges and universities in Illinois in the bill but said there'd be further discussions on whether the measure would be expanded to include those schools.
While he said he knows of no incidents of AI solely replacing classroom instruction, he explained 'that's the direction things may be moving' and that 'the level of experimentation with AI in the education space is significant.'
'I fully support using AI to supplement instruction and to provide students with tailored support. I think that's fantastic,' Rashid said. 'What we don't want is during a, for example, a budget crisis, or for cost-cutting measures, to start sacrificing the quality of education by replacing instructors with AI tools.'
While Keicher backed Morgan's mental health services AI bill, he opposed Rashid's community college bill, saying the language was 'overly broad.'
'I think it's too restrictive,' Keicher said. 'And I think it would prohibit our education institutions in the state of Illinois from being able to capitalize on the AI space to the benefit of the students that are coming through the pipeline because whether we like it or not, we've all seen the hologram teachers out there on the sci-fi shows that instruct our kids. At some point, 50 years, 100 years, that's going to be reality.'
Also on the education front, lawmakers advanced a measure that would help establish guidelines for elementary and high school teachers and school administrators on how to use AI. It passed 74-34 in the House before passing 56-0 in the Senate during the final hours of spring session.
According to the legislation, which has yet to be signed by Pritzker, the guidance should include explanations of basic artificial intelligence concepts, including machine learning, natural language processing, and computer vision; specific ways AI can be used in the classroom to inform teaching and learning practices 'while preserving the human relationships essential to effective teaching and learning'; and how schools can address technological bias and privacy issues.
John Sonnenberg, a former director of eLearning for the State Board of Education, said at a global level, AI is transforming education and, therefore, children should be prepared for learning about the integration of AI and human intelligence.
'We're kind of working toward, not only educating kids for their future but using that technology to help in that effort to personalize learning and do all the things in education we know we should be doing but up to this point and time we didn't have the technology and the support to do it affordably,' said Sonnenberg, who supported the legislation. 'And now we do.'