THE CONGRESSIONAL RESEARCH SERVICE AND THE FUTURE OF AI-ENABLED POLICY ANALYSIS

Committee on House Administration House June 25, 2026

Witnesses (1)

  • Karen Donfried, director, Congressional Research Service

From the witness list in the published hearing record.

Full text of the official published hearing record. Extracted from the source document — verify against the official record for citation.

- THE CONGRESSIONAL RESEARCH SERVICE AND THE FUTURE OF AI-ENABLED POLICY ANALYSIS

[House Hearing, 119 Congress]
[From the U.S. Government Publishing Office]

 THE CONGRESSIONAL RESEARCH SERVICE AND
 THE FUTURE OF AI-ENABLED POLICY ANALYSIS
=======================================================================

 HEARING

 BEFORE THE

 COMMITTEE ON HOUSE
 ADMINISTRATION

 HOUSE OF REPRESENTATIVES

 ONE HUNDRED NINETEENTH CONGRESS

 SECOND SESSION

 __________

 JUNE 25, 2026

 __________

 Printed for the use of the Committee on House Administration
 
[GRAPHIC NOT AVAILABLE IN TIFF FORMAT] 

 www.govinfo.gov
 www.cha.house.gov
 
 __________
 
 U.S. GOVERNMENT PUBLISHING OFFICE 
64-103 WASHINGTON : 2026
======================================================================= 
 
 COMMITTEE ON HOUSE ADMINISTRATION

 BRYAN STEIL, Wisconsin, Chairman

LAUREL LEE, Florida, Vice Chair JOSEPH MORELLE, New York,
BARRY LOUDERMILK, Georgia Ranking Member
H. MORGAN GRIFFITH, Virginia TERRI A. SEWELL, Alabama
GREG MURPHY, North Carolina NORMA TORRES, California
STEPHANIE BICE, Oklahoma JULIE JOHNSON, Texas
MARY MILLER, Illinois
MIKE CAREY, Ohio

 Mike Platt, Staff Director 
 Jamie Fleet, Minority Staff Director 
 
 
 C O N T E N T S

 ---------- 
 Page

 Opening Statements

Chairman Bryan Steil, Representative from the State of Wisconsin. 1
 Prepared statement of Chairman Bryan Steil................... 2
Ranking Member Joseph Morelle, Representative from the State of 
 New York....................................................... 14
 Prepared statement of Ranking Member Joseph Morelle.......... 16

 Witnesses

Karen Donfried, director, Congressional Research Service......... 3
 Prepared statement of Karen Donfried......................... 6

 Submissions for the Record

Washington Post article.......................................... 23

 Questions for the Record

Majority questions for the record................................ 45
Minority questions for the record................................ 48
Congressional Research Services answers to majority questions.... 52
Congressional Research Services answers to minority questions.... 69
CRS and OCIO IT Summary Table.................................... 99

 
THE CONGRESSIONAL RESEARCH SERVICE AND THE FUTURE OF AI-ENABLED POLICY 
 ANALYSIS

 ---------- 

 June 25, 2026

 Committee on House Administration,
 House of Representatives,
 Washington, D.C.
 The Committee met, pursuant to notice, at 10:21 a.m., in 
room 1310, Longworth House Office Building, Hon. Bryan Steil 
[Chairman of the Committee] presiding.
 Present: Representatives Steil, Lee, Griffith, Murphy, 
Bice, Morelle, Sewell, Torres, and Johnson.
 Staff present: Mike Platt, Staff Director; Abby Salter, 
Deputy General Counsel; Jordan Wilson, Director of Member 
Services; Jessica Smith, Deputy Director of Oversight; Libby 
Sippel, Deputy Director of Member Services; Sage Dobbs, Deputy 
Clerk; Jamie Fleet, Minority Staff Director; Khalil Abboud, 
Minority Deputy Staff Director; Matt Schlesinger, Minority 
Senior Counsel; and Kwame Newton, Minority Oversight Counsel.

 OPENING STATEMENT OF HON. BRYAN STEIL, CHAIRMAN OF THE 
 COMMITTEE ON HOUSE ADMINISTRATION, A U.S. REPRESENTATIVE FROM 
 WISCONSIN

 Chairman Steil. The Committee on House Administration will 
come to order.
 The title of today's hearing is ``The Congressional 
Research Service and the Future of AI-Enabled Policy 
Analysis.''
 I note a quorum is present.
 Without objection, the Chairman may declare a recess at any 
time.
 Also without objection, the hearing record will remain open 
for 5 legislative days so Members may submit any materials they 
wish to be included therein.
 We are here to discuss the Congressional Research Service 
and the critical role it plays in helping Congress carry out 
effective, informed oversight.
 The world is in the middle of a transformational moment. 
Rapid advances in artificial intelligence and other emerging 
technologies are changing how most sectors operate.
 Capitol Hill is no exception.
 Major technology companies, startups, and think tanks are 
rushing to develop tools that promise AI-enabled policy 
synthesis, legislative analysis, or new oversight capabilities.
 These efforts can have a net positive impact. Congressional 
staff are beginning to expect richer, more tailored analysis to 
support their work.
 In Members' offices, we are seeing something worth noting. 
Using the tools and process this Committee on House 
Administration has put in place over the last 3 years, staff 
are beginning to build their own internal tools to support 
their daily responsibilities.
 Let us go through an example of this.
 One example, Representative Keith Self's office. His team 
actually built an AI-based tool that provides summaries of 
introduced bills, outlines parliamentary consideration, 
recommends floor strategies, and even drafts a regulatory 
impact statement.
 Because they prioritize protecting taxpayer funds--it is 
Keith Self after all--they did this in a cost-effective manner, 
and it costs less than a dollar to run the analysis.
 This is an example of what happens when offices plan and 
execute new technologies. It shows that the legislative branch 
can be a leader in responsible technological adaptation. It 
offers early proof that AI can help Members carry out more 
effective oversight and legislative analysis.
 This is the landscape CRS finds itself in today, one that 
is changing quickly and where expectations are rising just as 
fast.
 CRS has an opportunity to complement and even shape this 
emerging ecosystem, bringing to it the values that have defined 
the service for decades: authoritative, trusted, and 
transparent analysis.
 Maintaining that role of thought leadership requires CRS to 
keep pace with the moment, even as it balances the significant 
technology investments it has made for the past decade.
 The question before us today is: How can CRS innovate 
boldly while remaining grounded in its core mission--serving 
every Member of Congress and every staffer impartially?
 How can CRS approach this challenge while remaining 
responsible stewards of taxpayer dollars, where every dollar 
invested in an IT system counts and where every CRS staff 
member is fully engaged?
 We look forward to hearing from CRS and from you, Doctor, 
about how you see this moment, how you understand the art of 
the possible, and how you plan to execute on a vision that 
allows the Service to step forward responsibly and confidently 
on behalf of Congress.
 [The prepared statement of Chairman Steil follows:]

 PREPARED STATEMENT OF CHAIRMAN OF THE COMMITTEE ON HOUSE 
 ADMINISTRATION BRYAN STEIL

 We are here to discuss the Congressional Research Service 
and the critical role it plays in helping Congress carry out 
effective, informed oversight.
 The world is in the middle of a transformational moment. 
Rapid advances in artificial intelligence and other emerging 
technologies are changing how most sectors operate.
 Capitol Hill is no exception.
 Major technology companies, startups, and think tanks are 
rushing to develop tools that promise AI enabled policy 
synthesis, legislative analysis, or new oversight capabilities.
 These efforts can have a net positive impact. Congressional 
staff are beginning to expect richer, more tailored analysis to 
support their work.
 In Members' offices, we are seeing something worth noting. 
Using the tools and process this Committee on House 
Administration has put in place over the last 3 years, staff 
are beginning to build their own internal tools to support 
their daily responsibilities.
 Let us go through an example of this.
 One example, Representative Keith Self's office. His team 
actually built an AI based tool that provides summaries of 
introduced bills, outlines parliamentary consideration, 
recommends floor strategies, and even drafts a regulatory 
impact statement.
 Because they prioritize protecting taxpayer funds--it is 
Keith Self after all--they did this in a cost effective manner, 
and it costs less than a dollar to run the analysis.
 This is an example of what happens when offices plan and 
execute new technologies. It shows that the legislative branch 
can be a leader in responsible technological adaptation. It 
offers early proof that AI can help Members carry out more 
effective oversight and legislative analysis.
 This is the landscape CRS finds itself in today, one that 
is changing quickly and where expectations are rising just as 
fast.
 CRS has an opportunity to complement and even shape this 
emerging ecosystem, bringing to it the values that have defined 
the service for decades: authoritative, trusted, and 
transparent analysis.
 Maintaining that role of thought leadership requires CRS to 
keep pace with the moment, even as it balances the significant 
technology investments it has made for the past decade.
 The question before us today is: How can CRS innovate 
boldly while remaining grounded in its core mission serving 
every Member of Congress and every staffer impartially?
 How can CRS approach this challenge while remaining 
responsible stewards of taxpayer dollars, where every dollar 
invested in an IT system counts and where every CRS staff 
member is fully engaged?
 We look forward to hearing from CRS and from you, Doctor, 
about how you see this moment, how you understand the art of 
the possible, and how you plan to execute on a vision that 
allows the Service to step forward responsibly and confidently 
on behalf of Congress.

 Chairman Steil. Ranking Member Morelle is tied up for a 
couple minutes. He is coming in. I am going to allow him to 
give his opening remarks in just a few minutes. All other 
Members will have until 5 p.m. today to submit their opening 
remarks.
 Today we have one witness. We have Dr. Karen Donfried, 
Director of the Congressional Research Service.
 Director Donfried, you are now recognized for 5 minutes to 
provide your opening statement.

 STATEMENT OF MS. KAREN DONFRIED, DIRECTOR, CONGRESSIONAL 
 RESEARCH SERVICE

 Ms. Donfried. Chairman Steil, Members of the Committee, 
thank you for this opportunity to discuss CRS and the future of 
AI-enabled policy analysis.
 AI is advancing rapidly, and CRS sees its potential to 
streamline our workflow and enhance our service to Congress.
 The use of AI, however, also carries risk--of outdated 
information, hallucinations, bias, and distortions.
 In this age of AI, I believe the role of CRS' highly 
skilled workforce will become even more important to Congress 
as you look for sources you can trust in an increasingly 
unreliable information landscape.
 I know the Committee will have valuable insights regarding 
how CRS can best support Congress by utilizing these new tools 
while preserving our five core values: of confidentiality, 
objectivity, nonpartisanship, authoritativeness, and 
timeliness.
 What has CRS' AI journey been like?
 In close partnership with the Library of Congress' Office 
of the Chief Information Officer, or OCIO, CRS has been 
actively exploring AI's potential.
 We are developing not only principles to govern the 
responsible integration of AI tools into our workflow, but also 
an implementation plan to identify priority use cases for 
development.
 In evaluating any AI-enhanced tool, CRS focuses on the 
extent to which these applications, one, increase productivity 
in developing products for Congress; and, two, align with CRS 
values.
 CRS has tested several AI applications. For example, we 
experimented with AI to produce bill summaries.
 CRS tested six different large language models over a 2-
year period. Approximately a thousand bills were fed into the 
models, yielding 3,000 AI-generated bill summaries.
 Less than 3 percent of the summaries met CRS standards for 
accuracy, coherence, relevance, and objectivity.
 While that result was disappointing, CRS has applied the 
lessons we learned to develop a plan for several use cases 
across the bill summary workflow that, if fully implemented, 
would increase our production of bill summaries.
 Our testing of AI applications has made clear the 
importance of training CRS staff. We have offered training on 
topics ranging from AI ethics to generative AI and prompt 
engineering.
 In fiscal 2027, in collaboration with OCIO, we are planning 
a 12-month evaluation to compare five AI tools: ChatGPT, 
Claude, Google AI, Perplexity and MS Copilot.
 As part of our fiscal 2027 budget proposal, CRS requested 
an increase of $1.6 million, including five permanent staff 
members with data science and AI development experience, which 
supplements the requests by OCIO to support the development of 
an AI platform that will be critical to our testing and 
deployment of AI applications.
 This platform would provide a dedicated and secure cloud 
environment to develop AI solutions tailored to CRS' unique 
requirements. This funding is essential to realize the most 
transformational impacts and productivity gains that AI offers.
 Along with understanding the promise of AI, CRS has gained 
valuable insight into its limitations.
 AI-enhanced technology excels at processing lots of 
information quickly, and, thus, can serve as a valuable tool to 
assist CRS experts.
 That said, many of the applications we tested produced 
results that fall well short of CRS standards for accuracy and 
objectivity.
 Quality control by CRS staff will be required to ensure 
that the end product is authoritative. Having that highly 
skilled human in the loop is essential to CRS building an AI-
augmented workflow consistent with our values.
 What is the bottom line?
 CRS exists for one purpose: to support Congress with 
objective, nonpartisan, authoritative, confidential, and timely 
research and analysis.
 Our most valuable assets in meeting that goal are the 
breadth and depth of expertise, knowledge of the legislative 
process, and institutional memory that our staff bring to bear 
on every issue of interest to Congress.
 AI increases the value of that high-quality CRS expertise.
 I appreciate the excellent relationship CRS has with you 
and with your staff and would like to thank you sincerely for 
your continued support of and trust in CRS.
 I welcome your questions.
 [The prepared statement of Ms. Donfried follows:]

 PREPARED STATEMENT OF MS. KAREN DONFRIED
[GRAPHICS NOT AVAILABLE IN TIFF FORMAT] 

 Chairman Steil. Thank you very much, Dr. Donfried.
 I will begin with questions, followed by the minority. I 
will recognize myself now for 5 minutes for the purpose of 
asking questions.
 Director Donfried, I will put this in about three different 
buckets of questions here in 5 minutes.
 CRS' authorizing statute guarantees complete research 
independence to CRS while requiring the most effective and 
efficient service to Congress.
 Balancing accurate--you brought this up a little bit in 
your comments about AI, kind of balancing the speed of 
response, the accuracy, the depth, how AI assists but also 
challenges this. You commented on the oversight and the 
accuracy of these AI models coming in.
 Can you provide some color as to how you and the team at 
CRS are thinking about leveraging this new technology while 
maintaining these standards and the speed at which you are able 
to deliver that product to Members and staff?
 Ms. Donfried. It is a great question, and I talked about 
the five core values that CRS has, and those values are really 
our promise to you.
 We are promising you what we send you will be nonpartisan, 
will be objective, will be authoritative, and we will do that 
research analysis in a confidential setting. No one is ever 
going to know what any of you asked CRS.
 We also pledge that we will get it to you in a timely 
fashion because we know that if you give us a deadline, if we 
get it to you after that deadline, it is not going to be 
useful.
 That is the promise. Yes, there is a tension across those 
values.
 When we say we want to give you something authoritative, 
let us say you give us a very short timeframe but you want a 
fully cited memo. We might call the staffer who put in that 
request and say, OK, we want to help, we know the timeline you 
have. We may not be able to give you that memo. What is the 
urgent thing that your boss needs to know, and we will do our 
best to get you that information.
 The promise is, in getting it to you in that timeframe, 
whatever we give you is going to be accurate. We always marry 
those values.
 Now, it is not going to be as deep an answer as if you gave 
us a week----
 Chairman Steil. Does AI--just to interject here--does AI 
speed that up in the sense of some of these AI tools can get 
you a quick, accurate answer, obviously, than in the current 
tools require human oversight to review, double-check, confirm 
those sources, although they may be able to pull up that source 
almost in real time.
 Ms. Donfried. The challenge that we are seeing with AI, 
when you come to research and analysis, is these chatbots are 
pulling from the internet. Some of them--Perplexity will give 
you cites. Many of them do not give you citations. We are 
having to verify any analysis that AI is producing.
 That is challenging. We are increasingly getting requests 
from congressional staff where they are saying ChatGPT gave me 
this answer. Is it true? Oftentimes, it is not accurate.
 Today we are not finding that that analytical or research 
function is speeding things up. Where it is really helping is 
it is very good at summarizing meetings and giving you meeting 
notes. It is very good at drafting emails. It is very good at 
coding. We can use those tools to provide code to put graphics 
into our reports.
 There are some things it is excelling at, and that saves us 
time and increases productivity, but we still think that 
verifying any of the research and analytical outputs is 
essential.
 Chairman Steil. How do you think about--I am going to 
totally shift topics for a final 90 seconds--siloing inside 
CRS? How do you build those collaborative teams where people 
have depth of expertise in certain areas, and you also have a 
separate IT team?
 How do you break through that in a period of time where 
technology may be able to level the playing field in that 
space?
 Ms. Donfried. Great question, and I would give you two 
answers to it.
 First, one of the real benefits of CRS is that you have 
people who have subject matter expertise and also functional 
expertise.
 Many of the complex policy issues you are working on, there 
is a policy element; there may also be a legal element, there 
may also be a question about legislative procedure to get that 
thing done. We can bring all of that to bear on that question.
 I actually think in that context we do a very good job of 
not siloing the research and analysis we are producing to you.
 The other piece of it relates to a realignment that I put 
in place last year directly connected to technology, because 
when I started as Director of CRS, I saw, in looking at the org 
chart, that there were people performing like functions in 
different parts of the organization. We created a new division 
that is focused on the role technology can play in the 
legislative work that we do.
 Chairman Steil. Cognizant of the time, thank you very much.
 I should have said in my opening--because I know there are 
a lot of men and women that work at CRS with you--we appreciate 
all their work as well. I know you work really hard, but I know 
there is a handful of people sitting behind you, but a lot of 
people at a building just next door that work incredibly hard 
on behalf of this institution. We appreciate all of their hard 
work.
 I now yield back. I will recognize the Ranking Member, Mr. 
Morelle, for a twofer, both for an opening statement, followed 
immediately by his questions.

OPENING STATEMENT OF HON. JOSEPH MORELLE, RANKING MEMBER OF THE 
 COMMITTEE ON HOUSE ADMINISTRATION, A U.S. REPRESENTATIVE FROM 
 NEW YORK

 Mr. Morelle. Well, thank you.
 I do apologize, Doctor and Mr. Chairman, for being late. 
Mrs. Bice and I were at the Congressional Art Competition 
thanking all the awardees. It got started a little late. Our 
apologies. Anyway, glad to be here.
 Thank you, Mr. Chairman, for convening this.
 Dr. Donfried, always a pleasure to see you and have you in 
front of the Committee today.
 For more than a century, Congress has relied on the CRS for 
accurate, authoritative, and nonpartisan information so we can 
better serve our constituents and carry out our constitutional 
obligations, responsibilities, and duties.
 To say CRS is busy would be an understatement. I am sure 
that you will be the last one to argue that point.
 In the most recent year for which we have public data--and 
this has probably already been disclosed, so I am repeating 
what you have already said--as I understand it, 75,000 custom 
responses to Member requests, 2,700 new and updated reports, 
6,500 bill summaries, 2,500 confidential memos, and 1,850 in-
person consultations. I feel like I am 1,800 of those 1,850. 
Thank you to you and your staff for all of that.
 The work, I understand, has not gotten easier. Like so many 
of their fellow legislative branch support offices, CRS has 
been asked to do more with less, even as the policy landscape 
grows more complex and the volume of legislation requiring 
analysis climbs year after year.
 Doctor, I want to recognize both you and your capable 
predecessor, now the Acting Librarian, Robert Newlen, for the 
steadiness with which you have brought the Service through a 
period of real transition. Thank you for that.
 I am looking forward and want to ensure CRS is making sound 
investments in the tools that will get the most out of the 
Service and its brilliant employees.
 I know in previous testimony you have said the adoption of 
any AI technology at CRS must enable adherence to CRS' core 
values and deliver on the potential to create efficiencies in 
the development and publication of products for Congress. I 
completely agree and appreciate that.
 I would like to underscore an important point. The most 
valuable part of CRS is not its technology, however, it is its 
people and the judgment they bring. The rigor, the expertise, 
the professionalism that lets your analysts and attorneys break 
down any question for any Member on any issue is why CRS is the 
gold standard.
 I will also say that we have been--and I have traveled with 
my colleagues different places to talk about the types of 
institutions that help other legislatures around the world, and 
it is clear you have no peer in that regard. You truly are the 
gold standard. I appreciate that.
 I will say, in this age of AI, the scale and speed of 
legislative work only make that standard more important, not 
less.
 I know--I see this--that I am sure Members will continue to 
use AI to draft legislation, which means the volume is going to 
go up, because instead of 2 or 3 weeks for staff to sometimes 
develop legislation, AI can do it in a matter of minutes.
 I am not sure that that is good necessarily, but that is 
the world we are living in, and we cannot make the mistake of 
thinking technology can be a substitute for a skilled and 
dedicated workforce.
 CRS investment in AI, as I am sure you agree, should make 
analysts more efficient, free them to do what only they can do, 
and they should never, ever replace them.
 I want to thank you again for your service, for being here 
today.
 Thank you, Mr. Chair, for holding this.
 [The prepared statement of Ranking Member Morelle follows:]

PREPARED STATEMENT OF RANKING MEMBER OF THE COMMITTEE ON HOUSE 
 ADMINISTRATION JOSEPH MORELLE

 Thank you, Mr. Chairman, for convening this.
 Dr. Donfried, always a pleasure to see you and have you in 
front of the Committee today.
 For more than a century, Congress has relied on the CRS for 
accurate, authoritative, and nonpartisan information so we can 
better serve our constituents and carry out our constitutional 
obligations, responsibilities, and duties.
 To say CRS is busy would be an understatement. I am sure 
that you will be the last one to argue that point.
 In the most recent year for which we have public data and 
this has probably already been disclosed, so I am repeating 
what you have already said as I understand it, 75,000 custom 
responses to Member requests, 2,700 new and updated reports, 
6,500 bill summaries, 2,500 confidential memos, and 1,850 in 
person consultations. I feel like I am 1,800 of those 1,850. 
Thank you to you and your staff for all of that.
 The work, I understand, has not gotten easier. Like so many 
of their fellow legislative branch support offices, CRS has 
been asked to do more with less, even as the policy landscape 
grows more complex and the volume of legislation requiring 
analysis climbs year after year.
 Doctor, I want to recognize both you and your capable 
predecessor, now the Acting Librarian, Robert Newlen, for the 
steadiness with which you have brought the Service through a 
period of real transition. Thank you for that.
 I am looking forward and want to ensure CRS is making sound 
investments in the tools that will get the most out of the 
Service and its brilliant employees.
 I know in previous testimony you have said the adoption of 
any AI technology at CRS must enable adherence to CRS' core 
values and deliver on the potential to create efficiencies in 
the development and publication of products for Congress. I 
completely agree and appreciate that.
 I would like to underscore an important point. The most 
valuable part of CRS is not its technology, however, it is its 
people and the judgment they bring. The rigor, the expertise, 
the professionalism that lets your analysts and attorneys break 
down any question for any Member on any issue is why CRS is the 
gold standard.
 I will also say that we have been--and I have traveled with 
my colleagues different places to talk about the types of 
institutions that help other legislatures around the world, and 
it is clear you have no peer in that regard. You truly are the 
gold standard. I appreciate that.
 I will say, in this age of AI, the scale and speed of 
legislative work only make that standard more important, not 
less.
 I know--I see this that I am sure Members will continue to 
use AI to draft legislation, which means the volume is going to 
go up, because instead of 2 or 3 weeks for staff to sometimes 
develop legislation, AI can do it in a matter of minutes.
 I am not sure that that is good necessarily, but that is 
the world we are living in, and we cannot make the mistake of 
thinking technology can be a substitute for a skilled and 
dedicated workforce.
 CRS investment in AI, as I am sure you agree, should make 
analysts more efficient, free them to do what only they can do, 
and they should never, ever replace them.
 I want to thank you again for your service, for being here 
today.
 Thank you, Mr. Chair, for holding this.

 Chairman Steil. You can continue.
 Mr. Morelle. I will go right into my questions. This is 
great. I get to reset the clock. That never happens.
 Maybe I can just, then, transition into--and I apologize if 
the Chair asked this, if I am asking to repeat anything--but 
talk about your experience using AI to produce bill summaries.
 What have you learned? What lessons should we all be 
mindful of as we not only work with you, but work with our own 
staffs and develop standards about the use of AI?
 Ms. Donfried. Thank you, Ranking Member Morelle.
 I just want to thank both of you for sharing your support 
of the 595 people I work with at the Congressional Research 
Service, because I am here representing them. I hope I will do 
it and make them proud. I am here to answer all your questions, 
but those are really the heroes of all of the work that comes 
out of CRS.
 Thank you for asking me about bill summaries. I think you 
were talking about how AI can produce bills at a really quick 
speed. Turns out you all can produce bills at a really quick 
speed. I think you are up to almost 17,000 in this Congress. We 
have a----
 Mr. Morelle. Which I will just pause, and I do not know if 
the other Members have this experience. When you are at an 
airport and someone says, what do you think about H.R. 6571? 
Like, I do not know.
 I appreciate you highlighting how many bills are 
introduced.
 Ms. Donfried. We have got 12 people who are summarizing 
those bills. Technology is our friend. We would love to figure 
out how technology can help us summarize that large number of 
bills.
 There was enthusiasm about an experiment to test the role 
of AI, and I shared in my opening comments that we were a bit 
disappointed because those six LLMs that we tested, out of the 
3,000 bill summaries that came out of it, there was an accuracy 
rate of 3 percent.
 It was really good at summarizing bills to rename a post 
office, but when you got to any level of complexity, it was 
really challenging for the LLM.
 Mr. Morelle. I am sorry. I misunderstood you. Say that 
again about the accuracy level.
 Ms. Donfried. We tested six large language models. We fed 
in a thousand bills. We got 3,000 bill summaries at the end of 
that process. When our analysts reviewed them for accuracy, 
objectivity, they assessed that there was a 3 percent accuracy 
rate.
 Mr. Morelle. 97 percent inaccurate?
 Ms. Donfried. 97 percent of those bills would never have 
met our standard for sharing them with you.
 Now, what are our lessons from that? I mean, one lesson, 
which was your question, was writing a bill summary is more 
complicated than we might think at first blush. Because you are 
not actually summarizing the bill. What we are doing is 
analyzing how that bill would change existing law, and it turns 
out that is a little more complicated.
 When those LLMs are drawing from the internet, they are 
coming up with some unusual things. OK. More complicated. OK. 
How could we use technology?
 Second lesson learned is, well, do not throw the baby out 
with the bathwater. Maybe AI was not great at producing those 
bill summaries, but let us break down the bill summary workflow 
and see if AI tools can help us along that process.
 We were able to use end-of-year money to work with a 
contractor, and we now have an implementation plan for discrete 
places in that workflow where AI could help us prioritize bills 
going to the floor. I will not go through all the examples.
 That is our next step of how we think AI can allow our 
analysts to be much more productive in producing the bill 
summaries.
 Third lesson learned is, unfortunately, you need resources 
to do that. That is connected to the Fiscal Year 2027 budget 
request. I realize it is a hard year. I realize there is not a 
lot of money. I am eyes wide open. We may not get that request.
 We will continue to move on that path, but we will be much 
slower and much more limited in what we can achieve.
 Thank you for the question.
 Mr. Morelle. I note your budget proposal for 2027 includes 
a request for five staff members with data science and AI 
development experience.
 Is that, too, in line with your request that you were just 
referencing?
 Ms. Donfried. Yes. When we talk about integrating AI tools 
into that bill summary workflow, we will need to customize 
tools for that.
 I work with really smart people at CRS, and many of them 
are integrating AI in important ways in the work that they are 
doing. We have a couple of examples. They have full-time jobs. 
They are not AI developers.
 We really need to buttress our workforce with, we think, 
five people is the reasonable estimate to achieve some of these 
goals.
 Mr. Morelle. You actually want them to help develop the AI 
tool? You are not going to outsource that? You think it is----
 Ms. Donfried. We have various uses for them. Yes, that 
would be one.
 Mr. Morelle. Thank you for all you do. Thank you and your 
team.
 With that, I yield back.
 Chairman Steil. The gentleman yields back.
 Representative Lee is recognized for 5 minutes.
 Ms. Lee. Thank you, Mr. Chairman.
 Thank you, Dr. Donfried, for being here with us today.
 The Congressional Research Service plays a unique and 
important role in helping Congress understand complex policy 
issues and serve the American people.
 As technology continues to transform how information is 
gathered and analyzed and shared, Congress must do our part to 
ensure that CRS remains both a trusted source of nonpartisan 
expertise and an institution prepared to meet the demands of 
this rapidly challenging environment.
 I appreciate your testimony about your human resources and 
the staff and expertise that you currently have--and also would 
like to have--to evolve with that environment.
 I am particularly interested in how emerging technologies, 
including AI, can help CRS better equip Members and staff to 
respond to constituents while still maintaining our necessary 
level of accuracy and rigor.
 On that note, we have had a great discussion already about 
the specific aspects of bill summaries and the potential use--
and, in fact, the unsatisfactory response--of the large 
language models in creating those bill summaries.
 Could you provide us with any specific examples where CRS 
is already using AI or advanced data tools to help analysts 
deliver information to congressional offices that have been 
successful?
 Ms. Donfried. Sure. Let me give you a couple.
 One I mentioned in passing, which is AI tools are proving 
excellent at coding.
 We found that our colleagues who are producing graphics--we 
have interactive graphics in many of our reports--we also have 
infographics. I brought you our AI taxonomy given the title of 
the hearing.
 These graphics, you need really quite a bit of coding to 
produce that result. We found AI is very good in speeding up 
that process and freeing up our visualization and graphics 
experts to do the work that AI cannot.
 I think these are helpful for you, for your staff, but also 
for your constituents. We have them on a whole range of issues, 
and graphics can be a great way to capture the nub of a complex 
issue. That is one example.
 Another example is, at congressional direction, CRS, 
starting in 2023, was asked to focus on data analytics. To 
date, my colleagues have created at this point five different 
data analytic models that we are using, everything from health 
insurance pricing to a student loan calculator.
 One of my colleagues has developed a tool that looks at the 
Regulations.gov website where they are pulling in comments from 
the public. This data analytical tool can look at those 
comments, synthesize and verify the results, and give you an 
understanding of how many comments are positive, neutral, 
negative.
 He is using an AI tool with that data analysis. He is 
continuing to develop it. It is another concrete example of 
where we are finding AI being very useful in helping us produce 
a policy-relevant result.
 Ms. Lee. You just touched on something that I think is also 
very important for each of our offices.
 We are increasingly asked to respond to constituent 
questions in real time, and they involve oftentimes many 
complex issues, some of which you just touched on.
 It could be anything from a specific policy area to 
disaster assistance or help with a Federal program or benefit, 
like veterans benefit or another Federal benefit.
 Are there uses that you have found in CRS--you just touched 
on a couple with the data analytics and the graphics--that you 
think would help our Member offices in responding quickly and 
accurately to those constituent needs? I am interested in the 
use cases you have identified and how you think those AI tools 
could be helping us serve our constituents better.
 Ms. Donfried. My advice to you today is go to CRS first. We 
are here to help you and your staff with that. We serve you in 
your legislative, in your oversight, and in your 
representational duties.
 We do a lot with your constituent requests, whether it is 
helping your staff understand what grants might be available. 
One of our most popular reports is actually a liaison for who 
can your staff contact in your district or in the Federal 
Government to help constituents.
 Because I am not confident in the accuracy of AI models to 
produce analysis for you, I would prefer you come to us, and we 
will do our very best to meet your needs.
 Ms. Lee. Thank you, Dr. Donfried.
 We are out of time. Mr. Chairman, I yield back.
 Chairman Steil. The gentlewoman yields.
 Representative Sewell is recognized for 5 minutes.
 Ms. Sewell. Thank you, Mr. Chairman.
 I wanted to thank you. I have used CRS a lot in my role as 
a Member of Congress, but also just to understand subject 
matter expertise in my Committee.
 I wanted to again echo what my colleagues have said about 
yourself and your amazing staff. As the Ranking Member was 
talking about you all being the gold standard, I just got back 
from the British Parliament, and I met with your counterpart in 
England, and they raved about how you all were the model that 
they wanted to follow.
 I just want you to know that CRS has really gotten a great 
reputation, not just within serving us, but with other 
parliaments. That is a great thing.
 Now, I wanted to know what you see as the future use of AI. 
You talked a lot about the study and only 3 percent was 
accurate. Are there--I mean, AI is improving daily, and there 
are so many different language models that are being tested.
 My question is: What do you see as the future use? Then 
what is the best use currently of AI? If not in analytics, is 
it in those graphics or is it--just explain a little bit more, 
if you do not mind.
 Ms. Donfried. Right. Our view of the future is, OK, we are 
going to integrate AI into the bill summary workflow, and then 
that is going to allow us to produce a much, much more 
significant number of bill summaries for you.
 We think that on the research and analysis side, it is how 
do we take advantage of the power of AI, the speed of AI, to 
process information that is more reliable, that can almost be 
like a research assistant to CRS analysts.
 That is going to take some time. I do not think that is a 
tomorrow thing. That is why we want to be testing these five 
different chatbots, because they all are a little bit different 
and excel at different things. It might be that we use X AI 
model for coding, Y AI model for that research assistance.
 We think there is a real opportunity there. First, we test 
it for our own use, but then we would imagine you all could 
benefit from it as well. If it is coming through us, you know 
it is meeting those standards of accuracy and objectivity.
 There are these very important use cases that we are moving 
toward, and we will have to see how quickly we move there. You 
are right, the technology is changing with every passing year.
 Ms. Sewell. I would think that there is a way that you 
could work with some of these models that are being tested by 
talking about the expertise and giving the analytics that is 
required so that the input into these AI models come back 
better quality. Is that even a possibility?
 Ms. Donfried. Well, if you look at--so I mentioned our 
budget request, which complements the Library's request. In an 
ideal world, we would like the Library of Congress to build an 
AI platform that we could train a large language model on.
 Ms. Sewell. Right, right.
 Ms. Donfried. That we would train it on legislative data.
 Ms. Sewell. Input your values.
 Ms. Donfried. Then you have circumscribed what it is 
pulling from. That would be fantastic.
 Yes, we have a lot of ideas about where we can go here, and 
I think it is transformational.
 Ms. Sewell. The best use of us, I think, is an 
appropriation. I wanted to know how much more you have asked 
that relates to CRS being technologically savvy and advanced.
 Ms. Donfried. Well, thank you for that.
 The Library has a request in. The Library has spoken to 
that.
 Ms. Sewell. Right.
 Ms. Donfried. The CRS request is for an additional $1.6 
million a year. That is primarily going to those five 
individuals I mentioned, data analytic and AI development 
experts.
 It also would allow us, when we are testing those five 
additional chatbots, to not only use the free models, but to 
have--I think, as we all know, there are multiple levels on 
these models, and the more sophisticated levels do perform 
better.
 It is for other things as well, but it is $1.6 million, and 
that is a recurring expense. We think that puts us in a very 
good position over the coming year, so we do not anticipate 
next year or the following year having an increase.
 Ms. Sewell. I think it also will save us money in the long 
run, I would think, in terms of quality, productivity, and the 
like.
 Thank you again for what you do for us and our 
constituents. Thanks.
 Chairman Steil. Mr. Griffith is recognized for 5 minutes.
 Mr. Griffith. Thank you very much.
 Let me join in the thanks. CRS has done a great job during 
my time here in Congress. It gets complex sometimes, and I 
greatly appreciate it.
 Ms. Sewell mentioned getting help on subject matter 
particularities or other stuff. Last year I had an experience 
where we had that.
 For the folks watching back home, you know it and we know 
it, but the committees up here do not always have clear-cut 
jurisdiction. Trying to figure out what the subject matter is 
of you are a particular committee or subcommittee can sometimes 
be interesting.
 I remember I was first here and there was an issue--I 
served on the Energy and Commerce Committee--and there was an 
issue about water outside of the mine, but that was affected by 
a mine in West Virginia. I am, like, oh, we have got to get on 
this right away, only to find out it was Natural Resources and 
not the committee that had primary jurisdiction over the EPA. 
If the water was in the mine, it was ours, but if it was 
outside on the ground, it was somebody else's.
 I took on a new task at the beginning of 2025 taking over 
the Chair of the Environment Subcommittee of Energy and 
Commerce. I am like, OK, what exactly can I go after, or what 
do I have to keep my hands off of because it is not mine?
 CRS did a great job of putting together a paper explaining 
all the jurisdictions and some of the history of the 
subcommittee. I am greatly appreciative for that.
 I am also appreciative for your skepticism on AI. Now, I 
love AI, and I think it has great potential, and I loved your 
example that you gave a minute ago about using it as a research 
assistant.
 In a timely article yesterday--and you probably saw it in 
The Washington Post--they did basically what you had done 
trying to figure out whether it would work for you, and it was 
only 3 percent effective on bill summaries. They went through, 
and they specifically asked political questions.
 A lot of times when I am using AI, I am asking, what year 
did a particular battle happen, if I am doing a history, or 
what was the time sequence on something, and it does pretty 
good on that.
 The causes of the conflict and so forth are more political 
in nature. The study--and, Mr. Chairman, if I might, I am going 
to offer the Washington Post article of yesterday by Kevin 
Schaul: ``Are ChatGPT and other AI chatbots politically biased? 
They tested them.'' The answer was yes.
 If you ask a political question, OpenAI came back 80 
percent of the time with a left-leaning argument and only 3 
percent of the time with a right-leaning argument. DeepSeek 
came back with 70 percent left-leaning and 7 percent right. Gab 
came back 50 percent left, 3 percent right.
 The rest of the percentages were in the middle. Anthropic, 
43 percent on the left, none on the right. xAI Grok came back 
40 percent on the left and 33 percent on the right. Google was 
7 percent on the left, but the rest of it was they presented 
both sides, which was kind of interesting.
 Now, whether they got it right or not, The Post did not go 
into. Of course, the companies all argue they did the test 
wrong.
 This was their study. They did quite an extensive study. I 
am going to ask that we submit that for the record in support 
of Dr. Donfried's statements.
 Chairman Steil. Without objection.
 [The article referred to follows:]
 [GRAPHICS NOT AVAILABLE IN TIFF FORMAT] 
 
 Mr. Griffith. We should be skeptical. It is a tool we can 
use.
 You also mentioned in your comments--and I just want you to 
talk about that a little bit more if you have time--that you 
want this extra money to train the AI. I think that is the real 
problem.
 If you have people--I do not think any of these companies 
set out to give left-leaning answers, but the people they hired 
to do the training of the AI were probably all leaning left, 
and to them this is the right answer. It was interesting that 
the Google one mostly came back with both sides, here is this 
side, here is that side.
 I would think that that would be helpful in your training 
and why you would need those five people. Am I correct in that?
 Ms. Donfried. Yes. Yes. I mean, I really appreciate--I was 
wondering if that article was going to come up about bias.
 It is so interesting--well, is it OK if I speak a little 
bit longer or are we out of time?
 Mr. Griffith. Yes. We have got 45 seconds.
 Ms. Donfried. I have talked to many Members of Congress who 
I would say are tech forward on both sides of the aisle, and 
one of them--we were talking about CRS and whether it will be 
replaced by AI--and he put that prompt into Perplexity. I got 
this two and a half page letter from him.
 I thought it was so interesting. It said the central 
challenge is not that AI will make research obsolete. It is 
that AI will make plausible-looking research cheap, fast, and 
ubiquitous.
 It is not that our eyes are closed and we do not want to 
embrace this technology. It is a tool. It is a powerful tool. 
It needs to be marrying that human expertise and experience 
with the technology.
 Mr. Griffith. To summarize, if you want C-minus to F work 
at this point in history, use the AI. If you want A work, which 
the CRS has always given us, you need the human factor, 
correct?
 Ms. Donfried. I appreciate that greatly.
 Mr. Griffith. Yield back.
 Chairman Steil. The gentleman yields back.
 Representative Torres is recognized for 5 minutes.
 Mrs. Torres. Thank you, Chairman.
 Thank you so much for being here, Dr. Donfried. Thank you 
for being here.
 I want you to also share my gratitude with your workforce. 
They do an incredible job on so many different topics at every 
level. I have used them constantly, and I find their work very 
useful.
 On the issue of the AI and how it could help us as Members 
of Congress be more transparent and more accountable, the 
Inland Empire, the area that I represent, is poor working class 
communities, some of the fastest-growing communities in the 
country during the good times. During the bad times we are the 
first ones to feel the hurt.
 Oftentimes my constituents are confused as to: Where do I 
go to find information about what Congress is voting on? How 
are they addressing the issues on the economy, the price of 
groceries, something so basic? What are you working on? How are 
you voting on the issues that are important to me?
 I think that that is a place that this Committee really has 
a huge responsibility to try to focus on providing some of that 
background work. I know the Chair of the Modernization 
Subcommittee, Rep. Lee, will be hosting a meeting next week, 
and I look forward to having you and getting more information 
on how we can modernize Congress more.
 I think a very simple tool that could help us be more 
transparent, for example, why are the vote tallies, how each 
Member is voting in the moment on the House floor, why isn't 
that public? Congress.gov can have amendments tacked up right 
away. Putting something on the website, I do not think that 
that is such a hurdle, so the American people know exactly what 
Congress is debating real time, not just the end result.
 Then, how our constituents are able to comment back, 
whether it is about a policy that Congress is reviewing. How 
are we measuring if it is bots that are providing feedback or 
if it is real humans providing that input?
 Those are some of the things that I am curious about, how 
you are working through some of that, and how you will be 
helping us modernize, be more transparent.
 Ms. Donfried. As I hear what you are saying, you are 
suggesting that Congress.gov could play an even larger role in 
leading to transparency of the workings of Congress.
 Mrs. Torres. Yes.
 Ms. Donfried. The Library of Congress is the owner of 
Congress.gov, and we curate and put certain data on 
Congress.gov, like the bill summaries and other things. I think 
what you may be referring to is I believe the deputy CIO is 
meeting next week to talk about Congress.gov, and I think he is 
very well-placed to talk about this.
 Congress.gov, we also work with many other stakeholders on 
the Hill, and I think there would be real interest in pursuing 
the ideas that you have.
 My understanding is that currently those, kind of the data 
about floor votes, goes on to Congress.gov I think within 24 
hours. It is likely that that could happen more quickly. It 
sounds like you are asking for that to happen maybe in real 
time.
 Mrs. Torres. Is it possible to have those votes that are 
happening real time online somewhere?
 Ms. Donfried. I, as a non-technologist, imagine it is 
possible, but I think that is something very much worth 
exploring.
 I am happy to take it back and share it with my colleagues 
in OCIO, and I think in that conversation next week, it is also 
a really good thing to explore.
 I know the Library and CRS are very open to responses about 
how helpful is Congress.gov in meeting your needs, and all of 
your constituents can pull up any CRS report.
 Mrs. Torres. Utilizing the AI to sift through those public 
comments to ensure that those comments are actually coming from 
real live people.
 Ms. Donfried. Again, I mean, I do think there is a way to--
an example I know, which may not be analogous, is there has 
been a huge spike in requests of Congress.gov, and I know OCIO 
is able to differentiate which of those are coming from 
machines and which are coming from humans. Overwhelmingly, it 
is coming from machines because a lot of folks are trying to 
scrape Congress.gov for the data.
 Mrs. Torres. I think we should be able to do it, too.
 Thank you. I yield back to the Chair.
 Chairman Steil. The gentlewoman yields back.
 Dr. Murphy is recognized for 5 minutes.
 Dr. Murphy. Thank you, Mr. Chairman.
 I need AI to turn on the microphone.
 Thank you, Mr. Chairman.
 Good to see you again. Guten Tag.
 Ms. Donfried. Good to see you. Guten Tag.
 Dr. Murphy. I am sitting here trying to think, which is 
always a challenge. I asked AI, I said, should Congress replace 
AI--should Congress replace the Congressional Research Service 
with AI? What did it say? It said probably not.
 Ms. Donfried. Whew.
 Dr. Murphy. Probably not. I know there is a little bit of 
angst and everything.
 This is really a fascinating conversation, just living in 
the world of medicine and just seeing the takeover of AI and 
how we are doing things so radically different, literally, in 
the last 3 years than we were just in the last 3 years and how 
the next 3 years are going to be an explosion.
 This is going to change our entire way of life, hopefully 
for the better. I think there are some--I do not want to say 
warmongers, but people who think the world is going to crash to 
an end.
 To my good friend, Representative Griffith, the point is, 
it is what we put into it.
 This is where you guys--I have to applaud, you guys are a 
nonpartisan service. You give us the data, you give us the 
facts, this is what it is therefore. The fact that if you do 
certain searches on certain things, you will get answers that 
have the proclivity to move in a different political direction.
 I am going to say probably not, I am going to say 
definitely no to AI when we are doing this.
 I think to the point of constituent services--I was just 
talking with Mrs. Bice next to us--I think it is a great idea, 
in part, because right now in the field of medicine we have AI. 
If a patient calls up with a particular problem or maybe a 
post-op thing, you have a bot, as it were, that will sit there 
with the patience of Job and answer every question.
 The problem is we get a lot of questions about the IRS, we 
get a lot of passport, things that are all confidential, and 
they are not able to be queried. I do not see that being an 
everyday part of this.
 We are going through this, I am sitting here thinking, what 
can I ask you? Because there are so many different points here.
 I would just hope that you are embracing this and not 
running from this because we are not going to be able to run 
from this. I had a fellow at a recent convention that said he 
did not want AI in his life at all. Not a bad thing to take a 
fishing pole and go to Montana, but we cannot really do this.
 As we move forward, what do you see the next 5 years 
looking like? How is it going to change your workflow? How are 
you going to adapt to it? Are you guys fully committed that 
this is happening and that CRS is going to have to change the 
way it does business, but also still be able to deliver the 
excellent care as you move forward?
 Ms. Donfried. Right. Thank you for the question. I hope 
there is nothing I have said this morning that suggests we are 
running away from AI. I just want to repeat, I think AI is a 
powerful technology that CRS wants to and needs to harness.
 The experimentation, exploration we have done to date has 
taught us a lot, both about the benefits AI has to bring, but 
also some of the limitations. I mean, my crystal ball is 
cloudy. I do not know how AI is going to change, improve over 
the next 3 years.
 What I want is for CRS to be experimenting with this tool 
and understanding the ways AI can best help us be more 
productive in the work that we do for all of you.
 Yes, we want to maintain those standards. Again, that is 
our promise to you, that we are giving you something 
authoritative, we are giving to you something that is not 
biased, that is nonpartisan, that is timely, and it is 
confidential.
 We want to meet all those markers. Of course, we want to 
harness technology. That allows us to be more effective in what 
we do for you.
 The things I have shared with other Members about what we 
have planned, could AI serve as a research assistant, how will 
this allow us to not only look at the bills that are going to 
the floor, which is how we are prioritizing things now, but 
look at, if not all 17,000, the majority of those bills.
 We have a lot of ideas about how we could use AI. If we can 
train an AI model on legislative data, that opens a whole new--
--
 Dr. Murphy. Right. I think, to your point--do not mean to 
cut you off--but to your point, you guys need to train the 
model, not take something off the shelf which is just 
inherently sterile, not the right thing, it is really going to 
be--it is going to be the best delivery of a product.
 Now, that will, obviously, take a lot of time, that will be 
intensive and require constant monitoring, constant adaptation 
as we embark upon this thing.
 It is a brave new world in a lot of, I think, good ways, 
but it is what we put into it. I appreciate the fact that you 
guys are going to be able to put in nonpartisan, just accurate 
information so that we can make our judgments.
 The last thing, good Lord, if we are putting out 15,000, 
17,000 bills, the last thing we need is AI to spit out another 
25,000. The Senate would love that.
 Ms. Donfried. We will need your help.
 [Sound of gavel knocking.]
 Dr. Murphy. Knocking noise.
 Ms. Donfried. To do this in real time and meet the 
challenge, we will need your support, and we will need some 
additional resources.
 Dr. Murphy. Great. Thank you.
 Ms. Donfried. Thank you.
 Dr. Murphy. I will yield back.
 Chairman Steil. The knocking is me. The gentleman yields 
back.
 Representative Johnson is recognized for 5 minutes.
 Ms. Johnson. Thank you, Mr. Chairman.
 It is such a treat to see you. I love CRS, you know that. I 
think you guys do amazing work and make it possible for us to 
be as thoughtful as we can in this important role that we all 
serve.
 Many of my colleagues have already asked a lot of the 
initial questions that I had, but I want to dig into kind of 
your comment about the 3 percent accuracy rate. I think that is 
alarming. I am stunned by that number. I bet that many of my 
colleagues up here are stunned by that number.
 I think it really highlights that we, as policymakers, 
cannot have a false sense of security of the accuracy of AI, 
and we have to really put guardrails on that and be vigilant 
about the data sourcing into these models.
 In your $1.6 million request, does that contemplate using 
publicly available models now or that does not include or 
contemplate you having your own model? Is that right?
 Ms. Donfried. Only if the Library portion of the request 
were also appropriated would we have the ability to train an 
LLM.
 We are complementing the Library's request, and our portion 
of it is specifically for five humans, allowing us to buy 
licenses for the more sophisticated level. It would need to be 
both of those in conjunction.
 Ms. Johnson. Yes. Because it is my understanding that 
really the AI models work, they cannot process all the data in 
the universe. They work at their best when they are narrowly 
tailored to a specific function or within a specific dataset or 
within a specific defined parameter of information and content 
that they can really perform at their best level.
 It is incumbent upon us to fund the Library's request in 
tandem with this to make sure that we can have our own internal 
AI model that is designed for the maximum productivity for CRS.
 Ms. Donfried. Two quick comments.
 I think if the Library were to develop that AI enterprise 
platform, it would serve the entire legislative branch. The 
Library is the repository of that data. I think the 
implications are far beyond the Library and CRS, A.
 B, the other important piece of this is, if CRS uses the 
free versions of these chatbots, of these AI models, those are 
not closed systems.
 We could not put any confidential material in there. We 
would not do that. We can only create a closed system with 
those higher levels of these AI models.
 Again, that is a pledge we have to you. We will always 
treat any congressional information in a confidential manner.
 That is where some of these resource questions become quite 
important.
 Ms. Johnson. Right. It seems troubling to me that the 
Government of the United States, the Congress, would rely on 
ChatGPT for all of our data.
 I joke with my mother. She is 85, and ChatGPT is her best 
friend. She puts a query into ChatGPT, and it says, oh, that is 
a brilliant question, you are so insightful. Then she will ask 
a follow up. That is a great clarification.
 It is so affirmatory of my mother's inquiries, and they may 
not be the best ones. They may not be the most clear, 
insightful question.
 I am becoming more skeptical of these giant AI structures 
that the Federal Government relies on for its data sourcing and 
information.
 I think, to your point that you have made in your 
testimony, there is no substitute for the breadth of the 
expertise of the CRS staff.
 It is just robust, it is impressive, the quality of 
expertise that you have there to bring to bear that, at least 
at this stage in AI development, it cannot be matched.
 I think you are right to assess the capabilities that AI 
can enhance your productivity, but certainly not replace it.
 I would be interested in seeing how Congress can help 
develop our own internal AI system, to your point of 
confidentiality in research and targeted usage. I think that 
would be an example of work this Committee should continue to 
pursue, and I think it would be very worthy going forward.
 Thank you so much for being here.
 Ms. Donfried. I appreciate it. I am going to use the 5 
seconds to say my favorite thing about CRS is all of my 
colleagues. My second favorite is new Member seminar.
 I know Congresswoman Johnson from new Member seminar, and 
it is something we do together with the Committee on House 
Administration, and I am really proud of what we do. I am 
looking forward to January.
 Ms. Johnson. Yes.
 Chairman Steil. The gentlewoman yields back.
 The gentlewoman from Oklahoma, the Chair of the 
Modernization Subcommittee, Representative Bice, is recognized 
for 5 minutes.
 Mrs. Bice. Thank you, Mr. Chairman.
 Yes. First let me say that I want to piggyback off of my 
colleague from Texas' comments. In your original statement, you 
mentioned that you had tested six LLM models to try to see if 
they were able to accurately summarize a bill, and that was 
they were not successful.
 I guess my first question to you would be, are we looking 
at potentially building a custom LLM that would be specific to 
CRS to be able to do bill summaries, or is that something that 
would be cost-prohibitive?
 Ms. Donfried. What we are looking at doing now is we are 
saying, OK, maybe AI cannot help us produce a bill summary out 
of the gate, and that is why we then said let us break down the 
bill summary workflow. What are all the discrete steps that our 
colleagues undertake to produce that bill summary?
 One of the first steps is right now we are prioritizing by 
saying: What are the bills that are going to the floor?
 We think an AI tool could be very helpful at doing that 
task and then telling the analyst: Here are the bills that are 
going to go to the floor, focus on those.
 Mrs. Bice. I guess what you are saying, though, is you are 
not looking at building an internal LLM and it is specific to 
congressional, sort of, language or bill language?
 Ms. Donfried. No. We may get there, but I am saying----
 Mrs. Bice. Right now we are not, though?
 Ms. Donfried. What we are trying to do right now is build 
that workflow tool that can allow us to produce more bill 
summaries.
 If we actually get funding, if the Library gets funding to 
build that AI platform and we can train an LLM on legislative 
data, it sets us up for then producing bill summaries.
 Mrs. Bice. I have enjoyed working with you over these last 
months and years. As the Chairwoman, I believe, of the 
Modernization and Innovation Subcommittee, along with the 
Ranking Member, Mrs. Torres, we have been able to really, I 
think, fine-tune the requests that we have coming from Members, 
but also to figure out how can we improve that interaction with 
CRS.
 In your Fiscal Year 2027 budget request, it includes $3.5 
million--a programmatic increase of $3.5 million to enhance AI 
and data analytic capabilities. You have signaled the funds 
would support expanded text and analytics tools, AI-enabled 
research, and broader IT modernization.
 My first question is, many external stakeholders have told 
us that there is a need for an API that will help them build 
new tools using public CRS information. Can you commit to 
making that tool available?
 Ms. Donfried. First, I just want to say that that increase 
in our appropriations ask is not all for AI. A good portion of 
it is for staff. 94 percent of CRS' budget is our people, and--
--
 Mrs. Bice. Let me stop and ask the question, though.
 Ms. Donfried. Yes.
 Mrs. Bice. Don't you think if we were able to utilize 
technology, you would not necessarily need additional staff?
 Ms. Donfried. Well, OK. First, let me say there is an API 
on Congress.gov. We have done that, and it is being used very 
actively. We have done that on Congress.gov.
 I am not ready to say that AI will allow us to have fewer 
staff. That may be right. What we are seeing today is AI is not 
doing the job CRS people are doing.
 If we talk about bill summaries, I do not think we are 
overstaffed on bill summaries. I think the 12 people we have 
writing those bill summaries, their challenge is there is no 
way they can keep up with the 17,000 bills.
 What AI is going to enable us to do is not have the 
incredible backload we have right now. I think it is enabling 
us to be more productive and, therefore, meet your needs 
effectively.
 It is not, at least today, obvious to me that we need fewer 
humans at CRS. We may get there. We are not that big. We are 
596 people. I believe the Ranking Member mentioned, they are 
producing 75,000 custom responses to you every year.
 Mrs. Bice. How far behind are you on bill summaries 
currently? Do you prioritize them?
 Ms. Donfried. The way we prioritize bill summaries now--and 
this was in consultation with you--was by saying we want to 
produce a bill summary on every bill that goes to the floor.
 Now we are meeting--97 percent of those bills that are 
going to the floor, we are getting you those bill summaries 24 
hours before they go to the floor.
 Yes, we have to prioritize. We do think AI will help us, 
obviously, create bill summaries that are not going to the 
floor. That is where we are today.
 Mrs. Bice. OK. Last quick question.
 Of the increase, how much is earmarked for concrete near-
term upgrades, like text analytics or workflow automation 
tools, that congressional staff can benefit from this year?
 Ms. Donfried. The 1.6 million that we are asking for for 
the AI part of the request----
 Mrs. Bice. That is 1.6 of the 3.5?
 Ms. Donfried. Exactly. Exactly.
 Mrs. Bice. The majority of the dollars are actually for 
additional staff.
 Ms. Donfried. Right, because, as you know, we have had a 
flat budget for the past 3 years. Our costs have gone up 8 
percent.
 The part of our budget that is not staff, the biggest part 
of it is research materials, which we have cut quite 
dramatically, but we cannot cut more because we need research 
materials to do that work.
 In any case, so, yes, the 1.6 million, that is primarily 
for those five humans with the special AI development and data 
analytical skills and to allow us to buy licenses for chatbots 
that are not the free versions. We are not in a place now where 
we have AI tools that we can give congressional staff to use. 
We would like to get there.
 Mrs. Bice. Mr. Chairman, thank you. I yield.
 Chairman Steil. The gentlewoman yields back.
 We appreciate your testimony today. I want to thank you for 
joining us. Appreciate all the work you and everyone at CRS 
does in supporting Congress.
 Members of the Committee may have some additional questions 
for you, and we ask that you please respond to those questions 
in writing.
 Without objection, each Member will have 5 legislative days 
to insert additional material into the record or revise and 
extend their remarks.
 If there is no further business, I want to thank the 
Members for their participation.
 Without objection, the Committee on House Administration 
stands adjourned.
 [Whereupon, at 11:22 a.m., the Committee was adjourned.]

 QUESTIONS FOR THE RECORD
[GRAPHICS NOT AVAILABLE IN TIFF FORMAT] 

 2024 Federal Employee Viewpoint Survey Responses From Congressional 
 Research Staff

 https://www.govinfo.gov/content/pkg/CHRG-119hhrg64103/pdf/
CHRG-119hhrg64103-add1.xlsx

 [all]