Print đź–¨

The Artificial Intelligence Advisory Council met on March 21, 2024, to hear invited testimony on AI systems in use by the Department of Information Resources, the Texas Department of Transportation, the Texas Workforce Commission, and the Teacher Retirement System.

 

This report is intended to give you an overview and highlight the various topics taken up. It is not a verbatim transcript of the discussions but is based upon what was audible or understandable to the observer.

 

Opening Comments

  • Sen. Tan Parker – Hearing from DIR, TxDOT, TWC, TRS; will be hearing through the year from many more state agencies & taking testimony from the public at large
  • Minutes from the last meeting and AI Advisory Council rules adopted; meetings will be held at call of co-chairs, presiding co-chair will alternate

 

Department of Information Resources

John Hoffman, DIR

  • DIR supports public sector industries in evolving tech like cloud services and AI; established AI Center of Excellence to educate and evangelize about AI, incl. prototyping of AI services, etc.
  • Hosted sessions and info on breaking tech silos, embracing bot usage, responsible AI usage in public sector, protecting privacy, etc.
  • Workshops include creating natural language processing engine, AI concierge, chatbots, data visualization, RPA, etc.
  • Initiated AI user group with 60 agencies represented, looking at bias, privacy, guardrails, etc.
  • AI User Group is great at sharing info; also held an AI Day covering these topics with agencies in attendance
  • AI Day 2 is set for May 31
  • Also providing AI training curriculum on training, ethics, transparency, etc.
  • DIR is considering AI efforts in lab environments as well, generate text & image summaries, VR

 

Jennie Hoelscher, DIR

  • Have been discussing agency concerns in the AI User Group; privacy is a significant concern, want to make sure any AI systems are protecting constituent data
  • Also concerned about potential inaccuracies of AI systems, stems from quality of data that models are trained on; agencies concerned that data used for training is not the best quality
  • Another risk identified is transparency and explainability of the models; want to be able to explain decisions & basis for decisions to constituencies
  • To mitigate risks, DIR is working towards finding way to implement governance and help agencies establish AI governance; User Group has a resource with all the governance documents being collected and created; many examples exist & different jurisdictions are doing different things
  • One of the most valuable is the NIST Risk Management Framework; helps agencies consider risk and mitigate; will likely hear of this from other agencies & could be something to consider adopting globally
  • DIR has a policy on the use of generative AI, any employee or team must submit request in writing and get approval for this; reviewed by info sec; a number of other agencies have adopted similar policies
  • Broad spectrum of what agencies are doing statewide, some are very restrictive, some are very open to generative AI in particular
  • DIR has been discussing if there is a need for AI rules statewide, AI use policy statewide, etc.; DIR has been collecting AI code of ethics from other jurisdictions and private entities
  • Some other states have said that systems in state government need to align with NIST RMF; includes a list of metrics for trustworthy AI systems
  • Dean Teffer, AI Advisory Council Member – On privacy and security, what is it about AI that creates larger risk?
    • Hoelscher, DIR – Potential for models to be trained on information and have public info remain in the model and be released down the road in some other format
    • Same privacy issues as other tech
  • Teffer – So generally is the solution not to train on PII and sensitive info?
    • Hoelscher – certainly the best way to do this, looking in included similar policies in contracts, etc.; statewide policy would be helpful
  • Amanda Crawford, DIR, AI Advisory Council Member – Can you talk about courses and training provided by DIR?
    • Hoffman, DIR – Have 135 licenses that were very popular, interest is overwhelming; over 600 hours of course training has been completed
  • Crawford – One of the biggest barriers to AI and other new tech is knowledge, re-skilling & upskilling
  • Sen. Parker – Who are the 135 licensees?
    • Hoffman – Ranges; generally technologists, but also licensees in business units
    • Can’t just train the tech people, must also train business, legal, procurement, etc.
  • Sen. Parker – Asks about benefits and movement in AI spaces
    • Hoffman – Since 2020, most of development has been in efficiency of the models; each agency has taken into account some AI opportunities within workforce, etc.
    • In our pilots, agencies are looking to maximize production of teams with limited resources
    • See this moving towards improvement of the constituent experience, ease of use
  • Sen. Parker – With regards to ethical guidelines, would like to see the guidelines collected by DIR; specifically concerning privacy, bias, etc., can you comment on that?
    • Hoelscher – Code of ethics are really focusing on things we can do to make sure there aren’t privacy issues, bias, etc.
    • Important to ensure that you have good data, employees need to understand the importance of good data and understand, storage, etc. of sensitive data/PII
  • Sen. Parker – Asks about largest concerns
    • Hoelscher – Biggest concern is quality of data, making sure systems are trained on good data, etc.; privacy is a concern, but without good data systems don’t work
  • Angela Wilkins, AI Advisory Council Member – Any applications that are more risky and we shouldn’t use AI?
    • Hoelscher – When looking at risk, need to look at data used and how sensitive the data is; systems using more sensitive data are higher risk
    • Also need to look at potential harm
  • Wilkins – How are you thinking about human in the loop to mitigate?
    • Hoelscher – Big component, any system significantly impacting Texans needs to have a human in the loop; component of codes of ethics in other jurisdictions
  • Rep. Capriglione – You had talked about making sure we can evaluate and explain decisions made by AI systems; how are you thinking about how to evaluate?
    • Hoelscher – Will be relying on 3rd parties a lot to develop these systems; up to developers and vendors to help us explain how the systems work
    • E.g. black box answer won’t be sufficient if it is a system for determining benefits to a constituent
  • Rep. Capriglione – Agrees, Council also needs to look at how state can assess and make appropriate decisions; how is the evaluation process going to work with vendors
    • Hoffman – Throughout process, need to put in all different components of security, ask how AI will be used, how will we have visibility & transparency, etc.
    • Once it gets to UAT, critical that it meets metrics & reqs, but also have to consider underlying data, accuracy, data generated, etc.
    • Look at similar things when we look at new cloud environments
  • Rep. Capriglione – Not fundamentally different with AI, still going to have software testing
    • Hoffman – One of the things we stress with teams, AI is a tech and process, controls, testing, etc. needs to remain in place
    • Might be more complex with new tech, but part of education & maturity of the organization in understanding these facets
  • Sen. Parker – What do you think this looks like 3, 5, 10 years from now?
    • Hoffman – Dangerous to speculate on tech; first focus is security & will be a key piece in how this progresses
    • Under constant threat today, defense tools we leverage currently use AI
    • Will need to look at how info is shared between gov sectors, etc.
    • Efficiency we will see in gov sectors will become commonplace; not necessarily a cost reduction, but reflects ability to meet growing needs
    • Some things are behind the scenes, e.g. emergency management scenarios; AI could be critical in response to certain events, predictions in demographic changes, needs for services, etc.
    • Will be a lot of obvious constituent experience things, but Ai will also drive these behind the scenes aspects
  • Sen. Parker – When you discuss this, two things come to mind; importance of accountability and transparency
    • Hoffman – Accountability is on different levels; each employee has a different level of accountability, but also important to understand accountability of vendors, etc. for systems they provide; IP could make it difficult for us to fully understand how systems are working
    • Look at AI as a great opportunity for transparency in gov, but need to be careful that some privacy consideration was not overlooked
    • Hoelscher – There is a higher burden in government to make sure what is deployed considers privacy, bias, constitutional protections, etc.
    • Can ask state information officer to come speak at a future meeting
  • Rep. Capriglione – Can’t fully use these systems without data to train on; what do we need to get this where it needs to be?
    • Hoffman – With data we’re a lot better off now than we have been; focus has been to identify most secure method and establish priorities in the data collection effort
    • Still room to go; all of the AI training efforts are matched on the data front
  • Wilkins – Could you teel us about some of the successes you’ve had
    • Hoffman – Big focus of DIR is cybersecurity, have AI validation embedded in auth system & this is used across different agencies
    • Other big effort is threat vector analysis & attacks on state resources
  • Sen. Parker – Have had big efforts over years to improve quality of data; important to have code of ethics that we follow, info from other entities will be useful; purpose of Council is to understand policy and needs, any low-hanging fruit the Council needs to address?
    • Hoelscher – There are some areas, could use some language designating who will write broad policy; other states have statewide policy either through legislation, executive order, etc.
    • Highlights Utah adopting AI Act this past week that creates a policy-making AI office; Utah is the only state that has gone down this path
  • Rep. Capriglione – How do you see the advent of cheap & accessible AI models and how it relates to cybersecurity? Hearing that it is easier for bad actors to develop viruses, use it for social engineering, etc.
    • Hoffman – It is a growing concern as we see these different threats; good news is that overall cybersecurity community is seeing it as well
    • Cybersecurity has been a tit-for-tat development effort for decades
    • Hoelscher – Key is sharing of info among cybersecurity orgs
  • Sen. Parker – Are there additional investments DIR needs to be able to manage AI and cyber threats?
    • Hoffman – Additional appropriations would be looked at favorably in tech and resources
    • Specifically around the AI component, interesting process given TX’s two-year cycle; agencies are currently prepping LARs for next session & should be bringing this up at AI Day, other forums
  • Wilkins – Sounds like you’re building a lot of the AI culture that will allow this to build into the overall system, DIR links orgs
    • Hoffman – We educate & enable, provide services for the public sector
    • We have a perspective across agencies, but each is responsible for their own tools & systems
  • Wilkins – Kind of a feedback loop also?
    • Hoffman – Correct
  • Rep. Capriglione – Many are using ChatGPT and some data is being put into this; how are you going to make decisions on what is usable in this versus other systems, knowing that systems will have a cost associated
    • Hoelscher – Some of this is related to risk management & how much risk is associated with the cost
    • Hoffman – Will need to make determinations on what is on-prem vs. cloud; will be interesting to see how governance is addressed at an agency level for different tech
  • Sen. Parker – What are we missing?
    • Hoffman – I think this is a good start, will continue to need to focus on tech partners & crafting tech
  • Sen. Parker – Are there things you need to do to ramp up dialogue with partners?
    • Hoffman- I think we’ve addressed them, but there is more work to be done
    • Needs to be part of conversations with vendors, need to expect a maturity with what is provided by partners

 

Texas Department of Transportation

Anh Selissen, CIO TxDOT

  • TxDOT wants to apply best practices with enterprise implementations; committed to innovation but also to safeguarding constituents and agency; governance needs to be sustainable and long-term
  • TxDOT was inventorying applications before HB 2060 & has established its own AI governance structure; similar to governance structure for other IT efforts
  • Person wanting to implement or develop system, AI or not, will go through governance process to review, determine accountability, agency impact, etc.
  • Have an AI risk framework and focus group to help understand tools & risk
  • TxDOT leadership is clear that AI shouldn’t be used for key decisions, should be used to back up
  • Have a strong community of practice to help understand potential risks and discuss specific projects
  • TxDOT is sending out regular training & education on AI, incl. safety; highlights education on tools like ChatGPT where data becomes public info
  • Highlights TxDOT’s previous experience in 2020 with a ransomware attack; have new practices in place of all IT functions, don’t want to create a separate track for AI
  • Highlights control policies, pen testing, impact assessments, etc.
  • Also have a process to look at blocked systems to follow up and review for usefulness
  • Firm believer that some of the accountability needs to be pushed onto the vendors, some reqs like TX-RAMP need to be more strongly enforced; vendors need to sign onto accountability policies
  • TxDOT wants to promote use of strong IT like AI, but also need to look at mitigating risk
  • Current AI initiatives include: 1) AI Incident Detection, pilot for traffic incidents on highways, 2) User Access Management, 3) Automated Invoice Processing, 4) Machine Learning for Video Analytics on Traffic Cameras
  • Seeing significant benefits from the incident detection pilot, incl. faster incident notification and response times and improved coverage
  • Increased workload also leads to increased need for manual processing, AI has been able to help get processing times down from 3 weeks to 24 seconds; using this for high frequency contracts & has led to a 45% reduction in manual labor
  • Future AI opportunities include: 1) PeopleSoft chat assistance, 2) smart fleet and asset management, and 3) crash & event prediction
  • Interested in creating a central enterprise data system with crash info to help perform analytics and modeling, also check for errors
  • Rep. Capriglione – Examples are compelling because they are “boring;” tools that drastically assist with backend processing of invoices, etc.; all of this translates to more efficiency & savings for taxpayers and is something we should be promoting
  • Rep. Capriglione – Were you surprised by how many applications TxDOT has using AI?
    • Selissen, TxDOT – Since ransomware in 2020, have been very proactive in looking at IT systems; weren’t surprised with the inventory of software as TxDOT has a detection system
    • There were some pieces where TxDOT needed to deep dive into what the software was and why it was used
    • Shock was more that some systems had been implemented by districts & TxDOT didn’t understand depth of what was being used
  • Rep. Capriglione – But helpful to go through this inventory process?
    • Selissen – Yes; understand power of AI, but it is like any other tech tool being implemented; strong foundation of security and controls meant we were already doing some of this when HB 2060 came out
  • Wilkins – Have you run into any areas you aren’t ready to use AI in?
    • Selissen – Productivity is where we’re doing a lot of this if it is low-risk, but also slow-step into other areas which is why TxDOT has a lot of pilots
  • Wilkins – Seems like a lot of vendors would be reaching out to you, how to you figure out which to work with?
    • Selissen – Yes; don’t have resources to implement every one, look at what is needed and what partnerships can be leveraged
    • Want to implement tools that will help fulfill mission of TxDOT and not necessarily IT
  • Teffer – On video capture for crashes, this has implications for privacy with facial geometry, etc.; what protections are there for that type of data
    • Selissen – Everything in the incident system is based on protection of data, have MFA on some of this & frequent vetting of security protocols
    • Sensitive to what is moved to EDP, no confidential info is moved into AI tools, etc.; don’t need private data to look at trends in incidents and traffic flow
  • Teffer – On future goals, utilizing AI for crash predictive analysis?
    • Selissen – Leadership is looking at this data in decision making for construction projects, etc.; tool isn’t making decision, but helping us focus
  • Sen. Parker – Efficiency examples are outstanding; are there additional lessons learned coming out of data your collecting, e.g. new patterns in traffic safety, etc.
    • Selissen – TxDOT has always been a huge proponent of data analytics, don’t necessarily think tools are providing new sources of info, but complimenting info to better justify construction projects
  • Sen. Parker – Asks about data usage and filtering, info sharing
    • Selissen – have steering committee, etc. looking at analytics, presented to leadership on a monthly basis and then filtered through over positions and workgroups
    • Have strong program making sure info is sent across TxDOT
  • Sen. Parker – Regarding collaboration partners; are you collaborating with outside parties, academic institutions, etc.?
    • Selissen – TxDOT is heavily involved with DOT conferences, big on sharing & collaboration; any time an AI tool is discussed in forums TxDOT looks at it, pilots it, etc.
  • Sen. Parker – What is the biggest challenge you’re concerned with and what have we not covered today?
    • Selissen – From an IT perspective, have changed how it is perceived over last 4 years at TxDOT, but a lot of overhead and costs are associated; funding needs to be looked at
    • Sometimes procurement process prohibits TxDOT from being agile and responding to markets; can sometimes go overboard in putting precautions in place where we can’t actually utilize tech to help productivity
  • Sen. Parker – Need to be able to move fast at times, can’t go through yearlong procurement on everything
  • Mark Stone, AI Advisory Council Member – With current AI initiatives, you are using offline initiatives, are you using multiple or just one? How were you able to procure this quickly?
    • Selissen – Some of the traffic monitoring is proprietary software via exemption through DIR
    • For automated tools for processing, it was a purchase off DIR listserv and procurement wasn’t necessary
    • Other items do need a procurement and this does take more time
    • For AWS and cloud tech, still building this up
  • Wilkins – Do you run into problems with sharing data with vendors, guidelines you follow?
    • Selissen – We are inputting data into software built for us, don’t share data with other public institutions, sensitive to data TxDOT has from data protection perspective
    • If we do put data in a vendor site, have strong language for protection and ability to get data back

 

Texas Workforce Commission

Heather Hall, TWC CIO

  • TWC has been an early adopter of AI for certain uses, e.g. in chatbot creation during COVID pandemic; chatbot was closely monitored for what questions it could and couldn’t ask, also have a tool called SARA working in vocational rehab program to help with scheduling, reminders, documentation, etc.
  • SARA lets staff prioritize customer engagements rather than perform admin duties
  • TWC’s Work in Texas system uses some algorithms to help pair jobseekers with jobs; next implementation will have new AI tools like helping employers write descriptions and jobseekers write resumes, etc.
  • Also use things like phone transcriptions, other products increasingly having AI tools built-in
  • TWC also put together an educational series for leadership, incl. ethics of usage
  • Highlights groups within the agency covering different areas & initiative with DIR to set up secure cloud data repositories where TWC can play with data and test AI tools
  • Developed a generative AI guidelines doc that is in alignment with existing cybersecurity manual
  • Established AI governance committee featuring all divisions within the agency
  • Require acceptance to deploy doc to be signed before deploying anything
  • Also seeing challenges, TWC is not used to be on cutting edge of tech & is a little uncomfortable for agency and customers; usually dealing with mature tech
  • Takes some time for new tech to get into governance cloud, working with DIR
  • Definitely found that tools aren’t ready to go out of the box; highlights HR manual being input into an AI tool and finding that questions are only answered correctly about half the time
  • Possible benefits are exciting, AI will not replace human staff

 

Adam Leonard, TWC

  • During pandemic saw a 14% increase in claimants, also an increase in bad actors trying to steal from the agency; TWC utilized machine learning to help detect fraudulent claims and cases; every fraudulent case feeds into data set to help detect future fraud
  • Using AI to monitor cases that could lead to a bad outcome & escalating those cases for more review
  • Another area is identifying those who are likely to be repeat filers to intervene and assist
  • Also could use tools to help match people with the programs or services that work best
  • AI will likely have a massive impact on jobs & this will mean many will need to make adjustments, reskilling, etc.
  • Developing a system for real-time monitoring of job postings and skills posted for, then projecting this into the future to feed info back to education and training providers
  • Sen. Parker – I love the focus on people, delighted to see progress made by TWC during COVID
  • Stone – Would very much appreciate analysis of changing job market, skillsets requested, etc.
  • Crawford – Echoes appreciation for focus on people; AI might not be the appropriate tool for a given gov interaction; also need to consider protections around data feeding models as it persists and models get more advanced over time; getting vendor community to understand this is a challenge, incl. contracting provisions the state couldn’t agree to
  • Crawford – Another key thing important to DIR is accessibility, are you seeing challenges with digital accessibility for AI tools?
    • Hall, TWC – Accessibility is important for TWC, vendor partners often struggle with complying with federal guidelines & often this doesn’t get us all the way on accessibility
    • With how fast AI tools are improving, worried that accessibility isn’t fully there; if it isn’t, we would not be able to use them due to customers and staff needing systems that work for them
  • Rep. Capriglione – Regarding testing and human oversight, how does this work? Are humans watching in real time, retrospective?
    • Hall – With chatbot during COVID, got a data dump every day of answer provided, humans assessed this every day for issues and fixes
  • Rep. Capriglione – This is key & whatever you can do to help other agencies with this would be good; we’re now using AI to write resumes, companies are using AI to filter, AI is assessing skills needed, etc.; at some point we need to step back and asak where humans are in the loop
    • Hall – With current tool, it is using resumes and cover letters written by individuals & still need to make choice to submit
  • Rep. Capriglione – Interesting to use analysis to see whether a person or case needs assistance; how do you determine analysis of sentiment?
    • Leonard, TWC – One of the key components is relationship between participant and the counselor, have extensive case notes that are taken by the counselors that can help a predictive model to help assess current cases
    • Have done a proof of concept & now looking at a model with more objective criteria
    • There are a lot of black box algorithms, but TWC has been discussing that anything the agency will do with potential to change what TWC recommends or pursues
    • Concerned about people reacting to the model and molding actions based on the model
  • Sen. Parker – Continuous monitoring should be across the board for any AI system, should be constantly making certain it has the outcome that was originally planned and tested; some functions at TWC are a good fit for AI like unemployment insurance, call vetting, etc., but need to maintain human touch
    • Hall – Focused on enhancing, not replacing
  • Sen. Parker – Success stories?
    • Leonard – Economy has gotten stronger and people are finding jobs more frequently
    • Not sure we’ve put into play yet a sufficient number of tools to determine outcome
    • Hall – COVID chatbot was a success story, answered large number of questions and tool was available 24/7; looking at AI tool that can supplement main door features and connect people to services
  • Sen. Parker – Asks about policymaking and workforce efforts
    • Leonard – As of two sessions ago the Texas Tri-Agency Workforce Commission was put into statute and given responsibilities
    • Recently launched MyTXCareer app to help people make better decisions on career options
    • Have an education outreach group to work with high school guidance counselors and others
    • Not doing a lot with AI in this space yet, but there are many compelling chat based approaches that can discuss options
  • Sen. Parker – You’re probably looking at building models to query job data from Comptroller and others & understanding needs
    • Leonard – Yes, part of National Labor Exchange which helps employers post job postings across the US, have data going back to 2010 that TWC is using to help project forward
  • Sen. Parker – What does AI do from a workforce perspective regarding helping older Texans find jobs?
    • Leonard – Anyone who hasn’t been in the job market for awhile would be shocked how it functions today and focus on skills and duties over titles; tools are designed to tease some of these functions out
    • AI can absolutely help with onboarding
  • Rep. Capriglione – Do you have ideas, stats, or studies on how many may lose their jobs because of AI? More in the camp that AI will result in cost-cutting and redundancies
    • Hall – Not aware of stats, but can take this back
    • Leonard – Have seen a number of high-profile news articles that companies like IBM have frozen hiring and are using attrition to bring down staff in certain areas, but don’t completely know impact; AI can also help people figure out how they can better do their jobs; most models show slight changes but also challenges with re-skilling
  • Sen. Parker – Wanted to focus on workforce changes, everything I’ve read is that there will be a wash or maybe some enhancement, but need to be mindful of things in the private sector; any other issues?
    • Hall – Do worry that adding additional layers will extend already long contracting process
  • Sen. Parker – Need to be able to meet speed things are happening so will need to find expedited procurement process
    • Hall – Also some grace going into session trying to figure out what some of these things cost
    • Leonard – People in their daily lives will be experiencing these changes, need to do these things carefully, but can’t afford to fall too far behind

 

Teacher Retirement System

Brian Guthrie, TRS Executive Director

  • TRS is in the crawling stages in the use of AI, have used machine learning tools for years, but taking very cautious approach with generative AI
  • Created AI policy in October
  • Need to make sure fiduciary duties are maintained for members to protect data, pay benefits, etc., but need to recognize that tech is evolving quickly
  • Exploring how to identify risk, evaluate portfolios and markets; important part is that we’re not planning to use AI to make decisions, but help people make decisions
  • Same with member interaction, don’t have a chatbot now, but plan to have a way to get basic info from the member and use that to arm whatever person members end up speaking to; could get to a point where it is an interactive chatbot, but nowhere near that currently
  • Have a set of data available to help identify trends in claims data, could eventually have a tool that can help do this
  • See potential, but haven’t implemented that yet
  • Highlights policy created in October, also have an AI Review Team with individuals from across the agency looking at tools on a case-by-case basis; haven’t rolled anything out yet
  • Part of review process is testing of products in a safe environment
  • Eventually do plan to roll out AI tools, some subject matter experts are excited about tools and they feel like they are being held back, but feel caution is necessary
  • Sen. Parker – I think you’re spot on with caution and your approach to the situation with the uniqueness of TRS and responsibility to pensioners, appreciate comment that AI won’t drive fund decisions; nothing wrong in TRS not having implemented AI given nature of agency
  • Rep. Capriglione – Agrees that TRS is unique, there is a difference in paying invoices and investing money; I’ve gone through your use policy and I think it’s great; in the policy it refers to employees wanting to use AI tech, does this also apply to an employee who wants to write an email or writing a computer program? How far does this go down
    • Heather Traeger, TRS – Applies to everyone, any use at TRS has to go through AI use policy, not permitted to use ChatGPT for example
    • Reviewing a number of different systems and possibility of running pilots
    • Process does require a use request, goes through manager, to AI Review Committee
    • Prohibition is broad to protect against issues like private data being uploaded to public data sets, etc.; Exec Director has ability to approve
    • Do want to get to place where smaller uses are approved, but concerned that info underlying any of these functions is TRS data
  • Rep. Capriglione – Can appreciate this, you have a pretty strict policy and repercussions can include termination
    • Traeger – Yes, will be evaluating and updating policy as appropriate
  • Wilkins – Are you looking at AI as a tool to pull data together for an overview?
    • Guthrie, TRS – We are looking at this, particularly with collating and summarizing investment data, a lot of which is public
  • Wilkins – You’re holding off at this point?
    • Guthrie – Have done this at a very high level, one of the use cases going through review committee right now
  • Sen. Parker – As you go forward what do you need? Anything specific to your agency from the work of this Council?
    • Guthrie – You’ve hit upon many of the issues everyone will be struggling with, coming up with fundamental reqs when developing a policy; every agency should have a policy
    • Privacy & security need to be prioritized, but one-size fits all isn’t appropriate given differing agency needs
    • Guidance from Council and legislature would be helpful to make sure we’re all obeying rules and not causing more harm than good
    • Trager – In noting that all agencies are different, there will be agencies that have to adhere to other regulatory regimes like HIPAA, etc., should consider when this occurs and is applicable

 

Closing Comments

  • Sen. Parker – Council does plan to take public testimony, notices for future meetings will be sent out as soon as they are available, now have a website up now where public can access past meetings