Federal Trade Commission Deloitte Complaint

FEDERAL TRADE COMMISSION 

Washington, DC 20580 

In the Matter of Deloitte Consulting LLP, a limited liability partnership, also d/b/a Deloitte 

 Complaint and Request for Investigation, Injunction, and Other Relief 

Submitted by National Health Law Program (NHeLP),  Electronic Privacy Information Center (EPIC), and Upturn, Inc.  

SUMMARY  

1. This complaint is brought to address serious financial and health harms to hundreds of thousands of Medicaid enrollees in Texas caused by a faulty Medicaid eligibility system developed by Deloitte Consulting LLP (“Deloitte”). In Texas, individuals seeking to enroll in Medicaid must submit information to—and have their Medicaid eligibility determined by—the Texas Integrated Eligibility Redesign System (“TIERS”), an automated benefits eligibility system developed and maintained by Deloitte. Although Deloitte has represented that its Medicaid eligibility system produces consistent and accurate eligibility determinations, hundreds of thousands of people have been and are being injured by the system’s failure to accurately automate the relevant eligibility rules. This results in loss of critical health care coverage and inability to access necessary medical care. 

2. Deloitte has been aware of similar accuracy issues with its automated benefits eligibility systems in other states for several years, but it has yet to take meaningful steps to mitigate the same harms to consumers in Texas who must rely on TIERS to access essential Medicaid benefits. The harms caused by Deloitte’s software are particularly acute at this time because all states are in the process of determining ongoing eligibility for millions of people who, by law, maintained Medicaid eligibility during the COVID-19 public health emergency. Deloitte has therefore failed to take reasonable measures to prevent foreseeable harm to consumers in Texas. Deloitte has also failed to show that its TIERS product is consistent with principles for responsible automated decision-making systems, such as those set out in Executive Order 14110 on the Safe, Secure, and Trustworthy Development and Use of AI and the White House’s Blueprint for an AI Bill of Rights.

3. Deloitte has engaged in unfair and deceptive trade practices, both directly and by providing the means and instrumentalities for unfair and deceptive trade practices, in violation of Section 5 of the Federal Trade Commission (FTC) Act. 

4. For the reasons set out below, the Commission should open an investigation, issue an injunction, and provide such other relief as proposed in Section VI, below, or as the Commission deems necessary and appropriate.

PARTIES 

5. The National Health Law Program (NHeLP) is a public interest law firm with offices in Washington, D.C., Los Angeles, California, and Chapel Hill, North Carolina. NHeLP was established in 1969 to protect and advance health rights of low-income and underserved people. NHeLP has been a leader in ensuring that automated decision-making and artificial intelligence (AI) do not harm Medicaid and other public benefit recipients, including serving as a founding member of the Benefits Tech Advocacy Hub.

6. The Electronic Privacy Information Center (EPIC) is a public interest research center in Washington, D.C., established in 1994 to focus public attention on emerging privacy and civil liberties issues. EPIC has played a leading role in developing FTC authority to address emerging privacy issues and to safeguard the privacy rights of consumers. EPIC is a longstanding advocate for the transparent, ethical, and responsible development, procurement, and use of algorithms and artificial intelligence.

7. Upturn is a non-profit organization in Washington, D.C., that advances equity and justice in the design, governance, and use of technology. Through research and advocacy, Upturn drives policy change by investigating specific ways that technology and automation shape people’s opportunities, particularly in historically disadvantaged communities. Upturn is a founding member, along with NHeLP, of the Benefits Tech Advocacy Hub.

8. Deloitte Consulting LLP is a private company having its principal Texas office at 500 West 2nd Street, Suite 1600, Austin, Texas 78701. Deloitte Consulting LLP develops, inter alia, software and other “technical and business solutions” (“systems”) to make Medicaid eligibility determinations.

9. The FTC is an independent agency of the United States government given statutory authority and responsibility by, inter alia, the FTC Act, 15 U.S.C. §§ 41–58. The Commission is charged with enforcing section 5(a) of the FTC Act, 15 U.S.C. § 45(a), which prohibits unfair and deceptive acts or practices in or affecting commerce.

 FACTUAL BACKGROUND  

A. Federal Requirements for Medicaid Eligibility Determinations. 

10. Title XIX of the Social Security Act establishes the Medicaid Act. The purpose of Medicaid is to enable each state, as far as practicable, “to furnish [] medical assistance” to individuals “whose income and resources are insufficient to meet the costs of necessary medical services” and to provide “rehabilitation and other services to help such families and individuals attain or retain capability for independence or self-care.”

11. State participation in Medicaid is voluntary. Participating states receive generous federal funding. All states participate.  

12. The Medicaid Act authorizes medical assistance coverage for designated low-income population groups and specifies health care services that states must or can furnish to covered groups.

13. In addition to fitting within a covered population group, an individual must have limited income and, for some population groups, limited assets. The income limits vary between groups and income eligibility is established using one of two sets of rules: (1) Modified Adjusted Gross Income (MAGI) rules, which count income based on federal tax rules and does not include an asset test, or (2) non-MAGI rules, which follow the Medicaid eligibility rules in place before implementation of the Affordable Care Act in 2014 and may include an asset test.

14. Medicaid coverage is a valuable benefit that enables low-income enrollees to purchase health care at little to no cost.

15. Individuals are enrolled in either “fee-for-service” Medicaid or “managed care.” Under the fee-for-service model, the state pays providers directly for each covered service received by a Medicaid enrollee, e.g., each time an individual goes to the doctor or fills a prescription. Under managed care, the state pays a fee to a managed care plan for each person enrolled in the plan (similar to paying a premium each month). In turn, the plan pays providers for all of the Medicaid services an enrollee may require that are included in the plan's contract with the state.

16. Individuals enrolled in Medicaid use their coverage to pay for many important products and services, including physician and hospital services, prescription drugs, medical equipment, and more.

17. Federal law requires that individuals with Medicaid coverage undergo renewal of their eligibility, also referred to as “redetermination,” at least every 12 months.

18. During the COVID-19 public health emergency, states paused annual Medicaid redeterminations in exchange for enhanced federal funding.

19. The Consolidated Appropriations Act of 2023 ended the continuous coverage requirement effective March 31, 2023. Accordingly, states re-started redeterminations in a process referred to as “unwinding.” 

20. When determining and redetermining Medicaid eligibility, states must “to the maximum extent practicable establish, verify, and update eligibility using . . . data matching . . . and determine . . . eligibility based on reliable, third party data.” 

21. To meet this requirement, each state must “develop . . . a secure, electronic interface allowing an exchange of data . . . that allows a determination of eligibility.”

22. This eligibility determination system must “streamlin[e]” the process for submitting initial applications and paperwork required for re-determinations of eligibility. And state Medicaid agencies must minimize the burden on individuals trying to obtain or retain eligibility.

23. A state system to support Medicaid eligibility and enrollment qualifies for a federal matching rate of 90% for design, development, and implementation and at 75% for ongoing operations if it meets certain requirements, including that it is “likely to provide more efficient, economical, and effective administration” and “support[s] accurate and timely processing and adjudications/eligibility determinations.”

24. Each redetermination must start with an attempt to determine the enrollee’s eligibility based on a review of reliable information already existing in state or federal records or available through a data exchange. This process is known as “ex parte” redetermination.

25. States have access to multiple databases that can verify eligibility during ex parte redetermination. For example, to check financial eligibility requirements, states can access a state wage or tax database, information from other means-tested programs like the Supplemental Nutrition Assistance Program (SNAP), or federal data from the Internal Revenue Service (IRS).

26. Information obtained through the ex parte process is then fed through the eligibility system’s logic, which automatically applies the program rules to make an automated eligibility determinations.

27. If the available information is sufficient to determine eligibility, Medicaid coverage will be renewed without requiring any information from the individual.

28. If an individual cannot be renewed ex parte, the state sends them a renewal packet to collect the missing information. The state may also request that the individual provide additional documentation to verify particular eligibility factors.

29. States may not ask for information that is irrelevant or that is available to it, and must not require individuals to provide information or documentation unless it “cannot be obtained electronically.

30. If an individual does not return the renewal packet or requested documents, Medicaid coverage is terminated. This is known as a “procedural termination” (as opposed to termination based upon a merits-based determination of ineligibility). 

 31. If the individual returns the requested information and documents, the additional information is populated into the eligibility software and fed through the eligibility system’s logic that automatically applies the program rules to make an automated eligibility determination.

 32. Before terminating Medicaid eligibility, a state must evaluate an enrollee’s eligibility under all categories of eligibility. To meet this requirement using an automated eligibility system, the system, must accurately evaluate for all categories of Medicaid eligibility.  

33. If the individual is not eligible under any Medicaid category, the state must evaluate their eligibility for other insurance affordability programs, including the Children’s Health Insurance Program (CHIP), and transfer the information to the other insurance affordability programs through its secure electronic interface.

34. The state must timely notify an individual of its decision regarding the renewal of Medicaid eligibility. The notice must, among other things, include a statement of what action the state is taking and the specific reason for the action. The notice must be in plain language and timely and be accessible to individuals with disabilities or limited English proficiency.

35. States must “provide such safeguards as may be necessary to assure that eligibility for care and services under the plan will be determined, and such care and services will be provided, in a manner consistent with simplicity of administration and the best interests of the recipients.” 

B. Texas and Several Other States Have Contracted with Deloitte to Automate Medicaid Eligibility Requirements. 

36. Deloitte contracts with at least 20 states to provide software to determine Medicaid eligibility.

37. These states include: Arkansas, Colorado, Connecticut, Delaware, Florida, Georgia, Illinois, Indiana, Kentucky, Michigan, Nevada, New Hampshire, New Mexico, Oregon, Pennsylvania, Rhode Island, Tennessee, Texas, Virginia, and Wisconsin.

38. Deloitte is one of, if not the, largest Medicaid eligibility determination software provider in the country.

39. In marketing materials, Deloitte describes itself as being able to give “consumers a single point of access across multiple health and human services programs” and having “the experience to meet federal requirements.” Deloitte touts its expertise in state and local systems, health care, and Medicaid systems specifically.

40. The Texas Department of Health and Human Services maintains a contract with Deloitte Consulting LLP to operate the Texas Integrated Eligibility Redesign System (TIERS) and an online public-facing portal, YourTexasBenefits.

41. Medicaid enrollees are individual users of Deloitte’s eligibility determination software, through both the online public-facing portal and mobile app, YourTexasBenefits, and through Deloitte’s automated decision-making logic in TIERS, which makes eligibility determinations based on enrollee data.

42. In states, such as Texas, where Deloitte holds a contract to provide the eligibility determination software, Medicaid applicants and enrollees do not have a choice to use an alternate provider.  

43. The online portal, YourTexasBenefits.com, is marketed as to enabling individuals to “submit redeterminations,” “check status of documents submitted,” and “check the status of their benefits.”

44. Deloitte represents to enrollees that “[t]he quick and easy way to send us proof is to use the website or mobile app. Uploading your files may help us review your case faster.”

45. Deloitte represents to Medicaid enrollees that “You can also go online to check the status of your case . . .  The 'Details' page will show you which benefits each person gets.”

46. TIERS is used to manage delivery of 64 “types of assistance, including food, cash, medical, and community care services to Texans in need. Over 10,000 unique users rely on TIERS every day.”

47. Individuals seeking to apply for Medicaid coverage can submit information to establish their eligibility through paper applications, over the phone, or through Deloitte’s online portal or mobile app, YourTexasBenefits. Regardless of how information is submitted, the data is entered into TIERS and individuals must have their eligibility determined using the TIERS system.  

 48. Texas is using TIERS to conduct Medicaid eligibility redeterminations for the more than six million individuals who will go through the Medicaid unwinding process by June 2024.

49. In its contract, Deloitte represents that TIERS “automates complex rule-based processing ensuring consistency in eligibility determination.”

50. TIERS “determines eligibility for [Health and Human Services] public assistance programs” including Medicaid.

51. TIERS “includes a central data repository to collect eligibility information,” a “decision table logic to process information and determine eligibility,” and software to generate client notices regarding eligibility determinations.

52. In Texas, the ex parte renewal process is also referred to as the administrative renewal process. 

53. The administrative renewal process begins with TIERS running an “automated renewal process.” “During the automated renewal process, TIERS checks for the required verification by program.”

54. “TIERS initiates the administrative renewal process without additional staff action. The administrative renewal process uses the automated renewal process to gather information from a person’s existing case and from electronic data sources to determine if the person remains eligible for Medical Programs.”

55. “The automated renewal process attempts to verify income by determining if the person’s income information is reasonably compatible with income information available through electronic data sources.”

56. “Once available verifications are assessed, the system runs eligibility.”

57. Beyond these generic descriptions, details of how TIERS is programmed to operate the automated renewal process are not publicly available. 

58. For individuals whose eligibility is not renewed through the automated process, TIERS generates the renewal packet and a list of documents to provide by a stated deadline and the packet and list of documents is mailed to the individual.

59. If the renewal form or requested documents are not provided by the deadline listed on the renewal packet  “the system automatically makes an eligibility determination through a mass update based on the eligibility outcome from the automated renewal process,” meaning the system automatically denies eligibility for failure to provide documentation. 

60. Deloitte, through the implementation of TIERS, is also responsible for generating a range of notices to Medicaid enrollees regarding their Medicaid coverage, including correspondence regarding how to maintain coverage and notices communicating final eligibility decisions. Deloitte was contracted to “develop and test English and Spanish versions of user interfaces, notifications, and correspondence for the TIERS Suite of Applications.”

61. For instance, if an individual is found ineligible, “TIERS provides 13 days advance notice to the household after informing them of a denial or termination of ongoing benefits using Form TF0001, Notice of Case Action.”

62. And if “additional information is needed and the client does not return a renewal form by the 30th day from the date Form H1211 is mailed, eligibility is auto-disposed and denied,” with no action from a case worker.

 C. Deloitte’s History of Inaccurate Eligibility Determination Systems. 

63. Benefits eligibility systems built by Deloitte have generated numerous errors, resulting in inaccurate Medicaid eligibility determinations and loss of Medicaid coverage for eligible individuals in many states.

64. These errors are ongoing and nationwide. A report on Deloitte’s “dysfunctional computer system” that repeatedly cut tens of thousands of New Mexico families from Medicaid coverage noted: “From Oregon to Rhode Island, state governments have shelled out billions of dollars to Deloitte for these so-called integrated eligibility systems, only to see those systems melt down, leaving poor families without a safety net.”

65. This is because “[o]ften states have rushed to implement the software before it’s ready—sometimes under pressure from Deloitte, according to workers in several state governments.”

66. For example, in Rhode Island, the Deloitte-built system was launched in 2016, despite concerns that it was not ready to go live. The system was plagued with numerous errors, including errors that inaccurately revoked individuals’ eligibility for the Medicare Premium Payment Program—and did so without providing the required advance notice of termination.

67. In Rhode Island, the “[t]he State and Deloitte intended to direct clients to apply online, but due to system issues many non-Medicaid applications that were submitted through the online portal since golive did not reach DHS workers.”

68. Furthermore, “[b]ecause of underlying data issues and Deloitte’s incorrect program rules (e.g., incorrect payment or benefit amounts), the system often fails to generate and issue accurate client notices on time. Notices provide basic information to providers and clients, and many are federally required.”

69. With respect to Medicaid, the Rhode Island system was also “unable to categorize Medicaid applicants into the correct sub-programs,” meaning the system did not correctly evaluate individuals for the various Medicaid eligibility groups.

70. Following the problems in Rhode Island, Deloitte acknowledged that it was aware of warnings that the system was not ready to go live and acknowledged that “[i]n retrospect it would have been a very good idea for us to go to pilot.” A report on Deloitte’s Rhode Island eligibility system concluded, “[a]lthough Deloitte was selected for its experience with these types of projects, it has not consistently adhered to industry best practices.”

71. Deloitte also launched the first phase of its integrated Medicaid and SNAP eligibility system in Illinois in 2013 despite known problems. And in 2017, following the launch of another phase after warnings the system was not ready, the resulting glitches caused long delays in access to essential benefits.

72. In New Mexico, when Deloitte’s integrated Medicaid and SNAP system was launched in 2013, individuals reported that the system was plagued with glitches, including that it would “swallow” documents people submitted to establish eligibility—i.e. the system would not recognize when documents had been submitted—resulting in a denial of benefits for failure to return the documents requested. 

73. The Deloitte system in New Mexico continued to experience glitches for many years. Reporting in 2019 described glitches that kicked foster children off of Medicaid and incorrectly denied health coverage to lawfully residing immigrants who are pregnant.

74. In Tennessee, when the Deloitte-built system launched in 2019, there were numerous errors in how it conducted eligibility determinations, including failure to deem newborn children Medicaid-eligible for 12 months as required by the Medicaid Act. Specifically, “there was a gap in the programming logic for when [the Deloitte-built Medicaid eligibility system] would link the report of a newborn’s birth to a mother,” and Tennessee “discovered that in some cases . . . TEDS [Tennessee Eligibility Determination System] was not automatically linking the newborn to her mother,” causing the newborns to experience gaps in Medicaid coverage.

75. In 2019, the Deloitte-built system in Tennessee also had errors in how it loaded data from the Social Security Administration’s State Data Exchange (SDX) database. As a result, individuals who should have been found eligible for Medicaid based on past receipt of Supplemental Security Income were found ineligible. Although the error was identified in 2019, it was not fully remedied until 2020, and individuals continued to lose coverage in the meantime.

76. In 2019, the Deloitte-built system in Tennessee also inappropriately found children ineligible for coverage during 2019 for failure to return verifications, even when it found parents in the same household eligible. The problem was not corrected until February 23, 2020.

77. In Pennsylvania, the Deloitte-built system could not conduct ex parte renewals for individuals who were also receiving SNAP benefits. Advocates trying to understand why the ex parte rates were so low ran into roadblocks because the processes and coding the eligibility system used to conduct ex parte renewals was not publicly available in the Medicaid handbook and was not produced in response to public records requests.

78. Deloitte system issues identified previously continue to appear in other states as well, wrongfully terminating coverage for eligible individuals, suggesting that Deloitte has not taken effective steps to ensure similar systems do not generate harm, even during the unwinding.  

79. For example, in Delaware, for several months during the 2023 unwinding, the Deloitte system generated notices that only provided individuals about two weeks to return the required renewal form when federal law and state policy require at least 30 days. Similar problems with inaccurate and missing system-generated dates and deadlines have also been noted in Colorado.

80. In Kentucky, during the 2023 unwinding, Deloitte’s system was reported to keep applicants in limbo. Advocates wrote to the State explaining that they had observed in the Deloitte system “a software error that is failing to flag to the system itself and workers that applicants have uploaded documents. As a result, the system appears to us to be automatically terminating Kentuckians’ health care coverage for ‘failure to submit documents’ that applicants have, in fact, provided,” and estimated that thousands of individuals had been impacted by this error each month.

81. In Arkansas during the 2023 unwinding, the Deloitte system experienced numerous glitches resulting in incorrect terminations. According to one news report “Three Medicaid eligibility workers—who were granted anonymity to speak freely about their work—told POLITICO that they have seen those glitches with the state’s new eligibility system, which was developed by the consulting firm Deloitte as part of a $340 million contract and launched in December 2020. The workers said they call them glitches because they seem to happen right after a system update is performed and because there’s nothing in clients’ files to explain the terminations. ‘They’ll do a mass update every so often and sometimes it accidentally closes the cases,’ said one eligibility worker, granted anonymity to speak openly about their work.”

82. During the unwinding, 13 states with Deloitte systems were out of compliance with the requirement that eligibility be renewed on an individual rather than a household basis. For the Deloitte states out of compliance, this issue affected at least 300,000 people (and with several states not providing an estimate at the time). This likely had a particular impact on children because they typically remain Medicaid-eligible even when their parents are not due to higher income eligibility thresholds.

D. Deloitte’s TIERS System is Faulty, Causing Hundreds of Thousands of People to Lose Medicaid Coverage and Deloitte Did Not Prevent these Foreseeable Harms. 

1. TIERS Does Not Reliably Conduct Ex Parte Redeterminations. 

83. TIERS does not reliably conduct ex parte redeterminations.  

84. Only 2.9% of individuals going through the renewal process in Texas have had their eligibility successfully renewed ex parte.

85. The national average for ex parte determinations is 32%.

86. Individuals who are not renewed ex parte must complete a renewal packet and potentially provide additional documentation. This substantially increases the risk that eligible Medicaid enrollees will experience a procedural termination for failure to provide documents.

87. For example, a recent news report documented a child with autism, Harper Wilson, whose Medicaid was terminated because TIERS determined that the family had not provided tax information. But tax information is readily available through federal data exchanges and should not be required where it can be obtained electronically.

88. After losing health care coverage, Harper’s family had to pay about $1,000 out of pocket for her home health aide each month. The family also had to forego Harper’s physical therapy appointments because they could not afford the weekly $150 cost. After missing several appointments, Harper’s health deteriorated and she had to visit the ER, moreover “Harper’s walking and balance are now three months behind where they might have been because she missed out on physical therapy, and she lost her spot at the facility while she was without coverage.”

2. TIERS Does Not Make Accurate Eligibility Determinations.  

89. Since unwinding began in Texas in April 2023, TIERS has repeatedly failed to make accurate eligibility determinations.  

90. Many of the Texas errors are the same as errors Deloitte’s software committed in other states and of which Deloitte was already aware.  

91. As in Tennessee in 2019, in 2023 in Texas, TIERS incorrectly terminated newborns from Medicaid coverage despite long-standing, mandatory federal eligibility rules establishing that newborns receive continuous eligibility for 12 months regardless of income or other changes. In Texas, workers noted in 2023 that “this newborn coverage issue is rooted in a distinct system glitch, poised to erroneously deny coverage for hundreds if not thousands of newborns.”

92. As in New Mexico in 2019, staff working with TIERS in 2023 have identified system errors affecting Medicaid coverage of eligible non-citizens and former foster youth.

93. As in Rhode Island in 2016, individuals in Texas in 2023 lost Medicaid coverage that would pay for Medicare Part B premiums and received no notice of their loss of coverage.

94. Texas staff have also recognized errors that have resulted in approximately 24,000 children losing coverage rather than being enrolled in CHIP.

95. Further, in April 2023, several thousand pregnant women lost coverage erroneously, though they were in fact eligible for coverage based on their pregnancies. In August, workers using TIERS wrote that “[t]he persistence of numerous and growing system issues remains evident, causing erroneous denials of coverage, particularly impacting newborns and pregnant women.”

96. Staff estimated that over 5,800 pregnant women “had their coverage denied and did not get the full two months of postpartum coverage.”

97. “The problems [are] so obvious that employees expressed frustration about the accuracy of the system.”Altogether, staff in Texas’s Health and Human Services Corporation working with TIERS have identified “over 20 active system issues, (and the list appears to grow weekly) each of which has either caused or is slated to cause disruptions in coverage.”

98. Attorneys working with families have also described repeated errors in eligibility determinations. As recently reported, “Terry Anstee, an attorney with Disability Rights Texas, said he’s seen situations in which the state determined people no longer qualified for Medicaid under one eligibility criterion and removed them, even though they were eligible under another. Other applicants appeared to have done everything right but were erroneously kicked off anyway. In a couple of confusing cases, Anstee said, people who qualified due to a disability were kicked off Medicaid and moved to the Healthy Texas Women program, which offers a completely different set of services. ‘It was just nonsensical,’ he said. ‘I don’t know if it was some sort of computer glitch—I really don’t know.’” 

99. Fixes to identified issues are often delayed and there is no indication that Texas has stopped redeterminations for the affected groups described above. For example, workers noted that in May “approximately 68,000 individuals erroneously lost coverage, only to be reinstated on August 5th, a span of nearly three months without coverage they were rightfully entitled to coverage.” Deloitte, therefore, has failed to take reasonable measures to prevent foreseeable harm to consumers who use TIERS to enroll in Medicaid coverage in Texas. 

100. Errors in TIERS also contribute to procedural terminations. As in New Mexico in 2013, Medicaid enrollees in Texas report that Deloitte’s online portal tells consumers they have no documentation to complete when they in fact do. These errors are recurring in Texas, despite Deloitte’s prior knowledge of the same errors in New Mexico.  

101. Medicaid enrollees report that, as in Kentucky during the unwinding, TIERS does not reliably recognize when individuals have submitted requested documentation. Enrollees who have submitted the requested information are nonetheless automatically terminated by TIERS and automatically sent a TIERS-generated a notice that they have lost eligibility for failing to provide that information.  

102. Medicaid enrollees report that Deloitte’s online portal does not present individuals with accurate information about their Medicaid eligibility status. For example, the online portal will often display inaccurate deadlines and dates of coverage. Enrollees who reasonably rely on the information presented in TIERS therefore, may miss deadlines to take action necessary to keep their coverage.  

103. Errors in TIERS’ eligibility determinations are routinely only identified after individuals have lost Medicaid coverage. Furthermore, repeating errors are routinely identified by individuals working with many enrollees, such as managed care plans, rather than by Deloitte.

104. Deloitte’s apparent failure to identify and mitigate similar harms already identified in other states and to use programs that do not accurately reflect long-standing federal mandates demonstrates that it does not maintain sufficient quality-assurance systems to detect errors and mitigate foreseeable harm to Medicaid enrollees.  

          I. LEGAL FRAMEWORK 

A. The FTC Act Prohibits Unfair and Deceptive Acts and Practices. 

105. Section 5 of the FTC Act prohibits unfair and deceptive acts and practices and empowers the Commission to enforce the Act’s prohibitions. 

106. The Commission has stated that a company violates Section 5 of the FTC Act not only when it directly engages in an unfair or deceptive trade practice, but also when it furnishes others with the means and instrumentalities for the commission of unfair and deceptive acts and practices.

107. An unfair practice is defined as one that “causes or is likely to cause substantial injury to consumers which is not reasonably avoidable by consumers themselves and not outweighed by countervailing benefits to consumers or to competition.” 

 108. Deceptive acts and practices include material representations, omissions, or practices that are likely to mislead a consumer acting reasonably in the circumstances.

B. The FTC Can Consider Established Public Policies Regarding AI and Automated Decision-Making to Determine Unfairness.  

 109. In determining whether a trade practice is unfair, the Commission may consider “established public policies.” 

 110. The OECD Artificial Intelligence Principles (“Principles”), the Blueprint for an AI Bill of Rights (“Blueprint”), the Universal Guidelines for Artificial Intelligence (“Guidelines”), Executive Order 14110 on the Safe, Secure, and Trustworthy Development and Use of AI, and the National Institute of Standards and Technology’s AI Risk Management Framework (“AI RMF”) each set forth established public policies for the use of artificial intelligence. 

 111. Medicaid eligibility determination software, including the software distributed by Deloitte, relies on data retrieved from third-party databases and a business rules engine to make automated eligibility decisions. This software is a form of automated decision-making that qualifies as AI under the Principles, Blueprint, Guidelines, Executive Order 14110, and AI RMF.  

             1. The Organization for Economic Cooperation and Development AI Principles. 

 112. The United States has expressly endorsed the Organization for Economic Cooperation and Development (OECD) Principles on Artificial Intelligence.

 113. OECD AI Principles define AI as machine-based systems that use input to “generate outputs such as predictions, content, recommendations or decisions.”Medicaid eligibility determination systems like those provided to states by Deloitte are AI as defined by the OECD AI Principles because they are machine-based systems that use input retrieved from third-party databases and Medicaid applicants and enrollees to make eligibility decisions.  

 114. Under the OECD AI Principle on Human-Centered Values and Fairness, “AI actors should respect the rule of law, human rights and democratic values, throughout the AI system lifecycle. These include freedom, dignity, and autonomy, privacy and data protection, non-discrimination and equality, diversity, fairness, social justice, and internationally recognized labour rights.” 

 115. According to the OECD AI Principle on Transparency and Explainability, AI actors should “provide meaningful information, appropriate to the context, and consistent with the state of art (i) to foster a general understanding of AI systems, (ii) to make stakeholders aware of their interactions with AI systems . . .  (iii) to enable those affected by an AI system to understand the outcome, and (iv) to enable those adversely affected by an AI system to challenge its outcome based on plain and easy-to-understand information on the factors, and the logic that served as the basis for the prediction, recommendation or decision.”

 116. The OECD AI Principle on Robustness, Security, and Safety states, “AI systems should be robust secure and safe throughout their entire lifecycle so that, in conditions of normal use, foreseeable use or misuse, or other adverse conditions, they function appropriately and do not pose unreasonable safety risk.” 

 117. According to the OECD AI Principle on Accountability, “[o]rganizations and individuals developing, deploying or operating AI systems should be held accountable for their proper functioning in line with the above principles.”

 118. The OECD Principles on Artificial Intelligence are “established public policies” within the meaning of the FTC Act.

2. The Blueprint for an AI Bill of Rights. 

119. The Blueprint for an AI Bill of Rights by the White House Office of Science and Technology Policy (OSTP) is a set of principles meant to guide the design, development, and deployment of automated systems in order to protect the rights of the American public.

120. The Blueprint for an AI Bill of Rights is “fully consistent” with the OECD AI Principles.

 121. The Blueprint applies to “(1) automated systems that (2) have the potential to meaningfully impact the American public’s rights, opportunities, or access to critical resources or services,” including “healthcare” and “social services.” The Blueprint expressly covers “systems that support decision-makers who adjudicate benefits such as collating or analyzing information or matching records,” and “systems which make benefits or services related decisions on a fully or partially autonomous basis (such as a determination to revoke benefits).” 

122. Medicaid eligibility determination software, like that provided by Deloitte, is therefore a type of automated system to which the Blueprint applies. 

123. According to the Blueprint’s Principle of Safe and Effective Systems, individuals should be protected from unsafe and ineffective systems. To that end, AI and automated systems should “undergo pre-deployment testing, risk identification and mitigation, and ongoing monitoring that demonstrate they are safe and effective based on their intended use, [and] mitigation of unsafe outcomes.” Further, the Blueprint states that “[i]ndependent evaluation and reporting that confirms that the system is safe and effective, including reporting of steps taken to mitigate potential harms, should be performed and the results made public whenever possible.” 

 124. The Blueprint’s Principle of Human Alternatives, Consideration, and Fallback recommends, “automated systems with an intended use within sensitive domains, including, . . . health, should . . .  provide meaningful access for oversight . . . and incorporate human consideration for adverse or high-risk decisions.” 

 125. The Blueprint’s Principle on Notice and Explanation recommends that systems should provide explanations that are technically valid, meaningful and useful to affected individuals and to any operators or others who need to understand the system, and calibrated to the level of risk based on the context.

 126. The Blueprint’s Principle on Notice and Explanation, in identifying why the principle is important, acknowledges that “[p]roviding notice has long been a standard practice, and in many cases is a legal requirement.”

 127. The Blueprint for an AI Bill of Rights are “established public policies” within the meaning of the FTC Act.

 3. The Universal Guidelines for Artificial Intelligence  

128. The Universal Guidelines for Artificial Intelligence (UGAI) are a framework for AI governance based on the protection of human rights that was set out at the 2018 meeting of the International Conference on Data Protection and Privacy Commissioners in Brussels, Belgium.

129. The Guidelines cover systems that include “some degree of automated decision-making” and “systems that impact the rights of people.”

130. Medicaid eligibility determination systems like those provided to states by Deloitte rely on automated decision-making and impact the rights of people to access health care. Therefore, they are artificial intelligence as defined by the Guidelines. 

131. According to the UGAI Right to Transparency, “[a]ll individuals have the right to know the basis of an AI decision that concerns them. This includes access to the factors, the logic, and techniques that produced the outcome.” 

132. According to the UGAI Assessment and Accountability Obligation, “[a]n AI system should be deployed only after an adequate evaluation of its purpose and objectives, its benefits, as well as its risks.” 

133. According to the UGAI Accuracy, Reliability, and Validity Obligations, “[i]nstitutions must ensure the accuracy, reliability, and validity of decisions.” 

134. The Universal Guidelines for Artificial Intelligence are “established public policies” within the meaning of the FTC Act. 

4. Executive Order 14110 on the Safe, Secure, and Trustworthy Development and Use of AI 

 135. On October 30, 2023, the White House published Executive Order 14110 setting out comprehensive guidelines to manage the development, procurement, and use of AI. 

136. Under Section 5.3 of Executive Order 14110, the White House encourages the FTC to “consider, as it deems appropriate, whether to exercise the Commission’s existing authorities, including its rulemaking authority under the Federal Trade Commission Act, 15 U.S.C. 41 et seq., to ensure fair competition in the AI marketplace and to ensure that consumers and workers are protected from harms that may be enabled by the use of AI.” 

137. The Executive Order states that the “Federal Government will enforce existing consumer protection laws and principles and enact appropriate safeguards” and “[s]uch protections are especially important in critical fields like healthcare…where mistakes by or misuse of AI could harm patients, cost consumers or small businesses, or jeopardize safety or rights.” 

138. Section 7.2 of the Executive Order also addressed the use of AI in Federal Government programs and benefits administration, including how states and localities use such systems in administering public benefits and how qualified recipients should be protected from unjust denials.

139. Section 2 of Executive Order 14110 sets out eight guiding principles and priorities concerning responsible AI development and use, including: 

A. Ensuring that AI is safe and secure through “robust, reliable, repeatable, and standardized evaluations of AI systems, as well as policies, institutions, and, as appropriate, other mechanisms to test, understand, and mitigate risks from these systems before they are put to use”; 

B. Ensuring that AI policies are consistent with the White House’s dedication to advancing equity and civil rights, including efforts to combat the “use of AI to disadvantage those who are already too often denied equal opportunity and justice” and to “hold those developing and deploying AI accountable to standards that protect against unlawful discrimination and abuse”;

C. Protecting the “interests of Americans who increasingly use, interact with, or purchase AI and AI-enabled products in their daily lives,” including efforts to “enforce existing consumer protection laws and principles and enact appropriate safeguards against fraud, unintended bias, discrimination, infringements on privacy, and other harms of AI”; and 

D. Protecting “American’s privacy and civil liberties,” including efforts to “ensure that the collection, use, and retention of data is lawful, is secure, and mitigates privacy and confidentiality risks.”

140. Executive Order 14110 is an “established public policy” within the meaning of the FTC Act.

5. National Institute of Standards and Technology AI Risk Management Framework. 

141. On January 26, 2023, the National Institute of Standards and Technology (“NIST”) published its AI Risk Management Framework (“AI RMF”), alongside various companion resources.

142. The AI RMF refers to “an engineered or machine-based system that can, for a given set of objectives, generate outputs such as predictions, recommendations, or decisions influencing real or virtual environments.” Automated Medicaid eligibility systems are AI as defined by AI RMF.  

143. The AI RMF is “designed to equip organizations and individuals… with approaches that increase the trustworthiness of AI systems, and to help foster the responsible design, development, deployment, and use of AI systems over time.” It is “intended to be practical, to adapt to the AI landscape as AI technologies continue to develop, and to be operationalized by organizations in varying degrees and capacities so society can benefit from AI while also being protected from its potential harms.” 

144. Under Section 5.1 of the AI RMF, NIST states that AI risk management processes and outcomes should be “established through transparent policies, procedures, and other controls based on organizational risk priorities” and that “organizational policies and practices [should be] in place to foster a critical thinking and safety-first mindset in the design, development, deployment, and uses of AI systems to minimize potential negative impacts.” 

145. Section 5.1 of the AI RMF also recommends that “organizational teams document the risks and potential impacts of the AI technology they design, develop, deploy, evaluate, and use, and they communicate about the impacts more broadly.”

146. Under Section 5.2 of the AI RMF, NIST states that organizations developing, selling, or using AI should examine and document the “potential costs, including non-monetary costs, which result from expected or realized AI errors or system functionality and trustworthiness.”

147. Under Section 5.3 of the AI RMF, NIST states that (1) “AI system performance or assurance criteria [should be] measured qualitatively or quantitatively and demonstrated for conditions similar to deployment setting(s),” (2) “the functionality and behavior of the AI system and its components… [should be] monitored when in production,” (3) “the AI system to be deployed [should be] demonstrated to be valid and reliable,” (4) “the AI system [should be] evaluated regularly for safety risks,” and (5) “[AI system] fairness and bias… [should be] evaluated and results [should be] documents.” 

148. Under Section 5.4 of the AI RMF, NIST recommends that (1) AI organizations should follow “procedures… to respond to and recover from a previously unknown risk when it is identified,” (2) “mechanisms are in place and applied, and responsibilities are assigned and understood, to supersede, disengage, or deactivate AI systems that demonstrate performance or outcomes inconsistent with intended use,” (3) “post-deployment deployment AI system monitoring plans are implemented, including mechanisms for capturing and evaluating input from users and other relevant AI actors, appeal and override, decommissioning, incident response, recovery, and change management,” and (4) “incidents and errors are communicated to relevant AI actors, including affected communities.” 

149. NIST’s AI RMF is an “established public policy” within the meaning of the FTC Act.

C. Deloitte’s Development, Maintenance, and Continued Deployment of the TIERS System Constitute Unfair and Deceptive Practices under the FTC Act. 

150. Deloitte’s inaccurate statements through its online portal, YourTexasBenefits.com, and its inaccurate eligibility determinations through the TIERS system are deceptive because they are material representations and omissions that mislead reasonably acting consumers. 

151. For example, as described above, Deloitte through the online portal, YourTexasBenefits.com, presents inaccurate deadlines, causing enrollees to miss deadlines to take action necessary to keep their Medicaid coverage.  

152. Deloitte also presents its inaccurate eligibility determinations to Medicaid enrollees through notices that are generated by TIERS. For example, TIERS also automatically generates notices stating required documents have not been provided. Where these notices are generated despite the Medicaid recipient having provided the information, the statements are deceptive.

153. And where TIERS inaccurately determines someone is not eligible for a category of Medicaid coverage due to a programming error, statements in the TIERS-generated notices communicating that ineligibility determination are likewise deceptive.  

154. Deloitte’s inaccurate eligibility information and determinations are reasonably relied upon by consumers to determine what actions they need to take, or not, to maintain their Medicaid coverage. The details of how TIERS makes eligibility decisions are not available to Medicaid enrollees whose eligibility is being determined. For example, a notice might state only that “You are not eligible for benefits” and “The money you get is more than allowed by program rules,” without describing what data sources TIERS relied on for income information, how TIERS calculated the household size and applicable income limit for that household size, or even that an automated decision-making system was used to make an eligibility determination. Thus individuals lack the information necessary to even identify an error in the TIERS eligibility decision. 

155. Accordingly, individuals reasonably assume that the TIERS eligibility decision is correct and may not challenge their loss of Medicaid coverage. As courts have recognized in the Medicaid context “there is a human tendency, even among those who are more experienced and knowledgeable in the ways of bureaucracies than the aged, blind, and disabled persons before us in this case, to assume that an action taken by a government agency in a pecuniary transaction is correct.”  

156. Deloitte’s inaccurate eligibility determinations and failure to correct known errors are also unfair under Section 5 of the FTC Act because they “cause[] or [are] likely to cause substantial injury to consumers which is not reasonably avoidable by consumers themselves and not outweighed by countervailing benefits to consumers or to competition.” 

157. Deloitte’s operation of its eligibility determination software systems is unfair because, as described above, the systems are deployed without sufficient measures to evaluate and mitigate risks of Medicaid coverage loss, directly injuring consumers. 

158. Inaccurate Medicaid eligibility determinations injure consumers in multiple ways including harms to health, financial harms, and the time, resources, and energy that must be spent to correct erroneous decisions. These types of “monetary . . . and unwarranted health and safety risks,” constitute substantial injuries.

159. Robust evidence shows that loss of Medicaid, even for short periods of time, leads to worse health outcomes, including premature mortality. These negative outcomes occur for a number of reasons. Churning on and off of coverage can result in skipping care and subsequent higher use of the emergency room, including for conditions like asthma and diabetes that can be managed in an outpatient setting. Even brief lapses in coverage increase the incidence of skipped medications and foregone treatment and result in worse health outcomes and increased use of the emergency department. Continuous insurance coverage is also associated with earlier cancer identification and outcomes. 

160. These harms are occurring in Texas. One recent news report described how “patients are showing up for appointments only to learn that they no longer have Medicaid and can’t afford to pay out of pocket to be seen. Even if their coverage is eventually restored, gaps can mean children missing vaccines or patients forgoing treatments for chronic diseases. Shanna Combs, president-elect of the Texas Association of Obstetricians and Gynecologists, said she’s heard about pregnant patients losing Medicaid just weeks before their due dates. At her practice in Fort Worth, she saw a child who was admitted to the hospital with a blockage in her lower genital tract, only learning then that she was no longer enrolled. ‘She basically has a uterus full of blood that’s probably up to her belly button’ and needed MRI imaging and ultimately surgery that the family couldn’t afford, said Combs. Doctors had to wait several weeks for her Medicaid to be reinstated, giving the child medication to try to prevent the condition from worsening in the meantime.” 

161. Studies also show that Medicaid coverage reduces medical debts and out-of-pocket expenses for enrollees and that loss of coverage imposes financial harms. That is because absent Medicaid coverage individuals must pay for necessary medical care, often at the expense of other basic needs or by incurring medical debt.

162. Medical debt is a major contributor to bankruptcies across the country. The financial benefits of Medicaid coverage have been repeatedly documented and have contributed to lower rates of bankruptcy. For instance, one study found that Medicaid coverage reduced the likelihood of borrowing money or skipping bills to pay for medical care by 40% and reduced the probability of having a medical debt collection by 25%. Without Medicaid coverage medical debt can be hard to control “unlike other sources of debt, people often fall into medical debt with total lack of consumer control. Seeking medical care differs from other consumer transactions since patients have little price transparency or provider options when receiving medically necessary services, especially in the case of an emergency.”

163. Evidence since the passage of the ACA also demonstrates how access to Medicaid in particular—rather than private insurance through the Marketplace or an employer—reduces medical debt and promotes financial security. For instance, one national study of low-income parents found that Medicaid expansion reduced difficulty paying medical bills and reduced stress and severe psychological distress. Ohio’s evaluation of its Medicaid expansion likewise reported substantial reductions in medical debt and improved ability to pay non-medical bills. And in Kentucky, evidence showed that among individuals who gained coverage under Medicaid expansion, average annual out-of-pocket spending decreased by $337, and the number of individuals who reported trouble paying medical bills decreased by 58%.  

164. Additional studies of the Medicaid expansion following the enactment of the Affordable Care ACt show significant improvements in financial well-being from Medicaid coverage. One study of credit report data found that when compared to low-income areas in non-expansion states, low-income areas in expansion states had significant reductions in unpaid non-medical bills and in the amount of non-medical debt sent to third-party collection agencies. Another national study found that medical debt fell by almost twice as much in expansion states (13%) compared to non-expansion states (7%). And a third study showed that Medicaid expansion reduced the incidence of newly-accrued medical debt by 30% to 40%, and also reduced the number of bankruptcies compared to non-expansion states. That study also examined the indirect consequences of unpaid medical debt, including reduced or higher-priced access to credit markets, and found that following expansion, credit scores improved significantly. 

165. Medicaid enrollees also suffer a “time tax” because of TIERS unreliable and inaccurate determinations. As described by the Office of Information and Regulatory Affairs, the time tax includes “the time, money, and psychological costs involved in interacting with government,” which puts a “tax on the time and well-being of individuals seeking assistance.” These burdens are exacerbated by “confusing notices, complicated questions, and underlying it all, the deep anxiety of potentially losing life-saving assistance.”  

166. The TIERS system’s low rates of ex parte renewals cause a substantial injury to Medicaid enrollees by doing “a small harm to a large number of people” and by raising “a significant risk of concrete harm." The dramatically low rates of ex parte renewals means that vastly more Medicaid recipients have to complete complicated forms and collect and submit documents to maintain Medicaid coverage, imposing significant administrative burdens on those enrollees.  

167. Furthermore, the low rates of ex parte renewals increase the risk that eligible individuals will lose Medicaid coverage for not providing necessary forms, including as for example, children like Harper Wilson, discussed above. 

168. Inaccurate eligibility determinations also impose a time tax because they require individuals to spend substantial time and resources to correct errors—either through informal efforts or through an appeal of the Medicaid eligibility decision through a “fair hearing.” 

169. To correct a decision informally, individuals may, for example, have to re-collect and submit documentation previously submitted. They often have to spend time on the phone to reach the call center, which in Texas can take a substantial amount of time. Particularly for families “where caregivers work in hourly jobs. Being on the phone is really difficult . . . because it could result in less pay.”

170. This “time tax” constitutes both economic and quantifiable non-economic harm, which establish “substantial injury.”   

171. And while individuals have the option to appeal the decision through a fair hearing process, that process can itself be onerous and time consuming, particularly for individuals with disabilities.

172. The burdens of correcting an erroneous decision are exacerbated because the information regarding how Deloitte developed and programmed the TIERS system to generate eligibility determinations in any particular case is not made available to Medicaid enrollees.  

173. Individuals do receive a TIERS-generated notice regarding the ultimate eligibility decision. However, this notice does not describe what data sources the TIERS system relied on or explain the logic for how TIERS reached its conclusion.  

174. As described above, a notice might state only that “You are not eligible for benefits” and “The money you get is more than allowed by program rules,” without describing what data sources TIERS relied on for income information, how TIERS calculated the household size and applicable income limit for that household size, or even that an automated decision-making system was used to make an eligibility determination.

175. Without that information, an individual cannot determine if TIERS, for example, calculated their income incorrectly or applied the wrong income standard, making it difficult to determine whether to try to challenge a decision. Many individuals may simply assume the system was correct about their ineligibility, because, without the specific information, they cannot tell that the system made incorrect decisions.

176. For individuals who do seek to challenge the decision, the lack of information requires individuals to expend more time and effort to first attempt to understand the basis of an erroneous decision before they can seek to correct it. Individuals are “force[d] into an unreasonable choice between either attempting to learn the reasons for the proposed action and then decide whether to appeal, or to appeal without knowing whether an appeal might have merit.”

177. The risks of appealing are even greater in Texas, where as the notices warn, individuals may have to repay benefits if they lose at the hearing. Thus, individuals must risk financial penalties to correct errors made by TIERS. 

178. Ultimately, the lack of information, confusion, and administrative hurdles can dissuade individuals from seeking to challenge the erroneous determination or trying to re-enroll. For example, one survey of parents revealed that the perceived hassle of applying, the complexity of rules and regulations, and confusion about how to apply were all significant factors that prevented parents from trying to enroll their children in Medicaid.  

179. As one parent in Texas recently explained “I hate to say this, but I wholeheartedly believe it: I feel that Medicaid just says no so parents will quit, and they almost capitalize on parents being tired and worn down by this whole process.” 

180. Deloitte’s inaccurate eligibility systems are likely to substantially impact a large class of people, including the six million Medicaid enrollees undergoing annual redetermination in Texas.  

181. The injuries caused by Deloitte’s inaccurate eligibility determinations cannot be reasonably avoided. Individuals are legally required to complete the annual redetermination process to maintain Medicaid coverage and must have their eligibility determination processed through TIERS. If TIERS determines an individual is ineligible, an individual cannot use their Medicaid coverage to purchase medical care until the decision in TIERS is updated.  

182. The harms caused by Deloitte’s inaccurate eligibility determinations are not outweighed by countervailing benefits to consumers or to competition. As described above, the harms to Medicaid enrollees who receive inaccurate determinations are substantial.  

183. There are no countervailing benefits to consumers or the State Medicaid agency from continuing to deploy an unreliable and inaccurate eligibility determination system. The Texas Medicaid agency is required to operate an eligibility system that performs enrollment functions and simplifies enrollment. Thus, any benefits to the Medicaid agency or state generally from the efficiency and automation of eligibility decisions are benefits from any system that would be implemented to comply with the Medicaid Act.  

184. On the other hand, there are no countervailing benefits flowing from Deloitte’s apparent failure to assess its eligibility determination systems for known errors that have been identified in other states. Deloitte could implement certain safeguards at a reasonable cost and expenditure of resources, such as inspecting and testing all of its Medicaid eligibility software for errors identified in one state. Thus, there are no benefits of Deloitte’s practice that outweigh the harm to Texas Medicaid enrollees. 

185. Deloitte is therefore directly engaged in an unfair trade practice in violation of the Federal Trade Commission Act, 15 U.S.C. §§ 45(a)(1). 

186. Alternatively, Deloitte furnishes state Medicaid agencies with the means and instrumentalities for unfair and deceptive acts and practices when it provides these agencies with eligibility determination software that it claims accurately and consistently automates Medicaid eligibility rules. Based on faulty eligibility determinations made by Deloitte’s TIERS software, Texas’s Medicaid agency has reduced or revoked public benefits from eligible consumers, causing the substantial harms outlined above, which cannot be reasonable avoided.  

D. Deloitte’s Practices Are Grossly Inconsistent with Established Public Policies Regarding the Development and Maintenance of Automated Decision-Making Systems. 

187. On information and belief, Deloitte does not follow key responsible AI risk management practices outlined in established public policies like the Principles, Blueprint, Guidelines, Executive Order 14110, and the AI RMF when developing and deploying automated benefits systems like the TIERS system, injecting unnecessary and serious risks of financial, health, and other consumer harms on benefits recipients. 

188. Deloitte does not provide meaningful information to foster an understanding of the determinations made by its eligibility systems that allow individuals affected to understand the outcome and challenge its outcome based on plain and easy-to-understand information on the factors and logic that was the basis of the decision in violation of the OECD Principle on Transparency and Explainability. 

189. Without meaningful information about the existence, role, and accuracy of the TIERS system within the Medicaid eligibility process, enrollees lack the knowledge they need to take steps to avoid harm caused by errors within the TIERS system. 

190. Deloitte has not ensured that its eligibility determination systems function properly and do not pose unreasonable safety risk throughout its entire lifecycle and in conditions of normal use in violation of the OECD Principle on Robustness, Security, and Safety, thus increasing the risk of substantial consumer injury while undermining any claims that the Deloitte system benefits consumers.  

191. The repetition of the same errors in Deloitte eligibility systems across Texas and other states and over time demonstrates that Deloitte has failed to perform the necessary testing, risk identification and mitigation, and ongoing monitoring to ensure safe and effective systems in violation of the Blueprint’s Principle of Safe and Effective Systems.  

192. Its failure to mitigate risks identified in other states also violates Section 5.4 of the AI RMF principles which directs organizations to follow “procedures… to respond to and recover from a previously unknown risk when it is identified.” Deloitte has not “deactivate[d]” Medicaid eligibility systems when they “demonstrate performance or outcomes inconsistent with intended use.” Nor does Deloitte report publicly on errors that have been identified, contrary to the recommendation to communicate “incidents and errors . . . to relevant AI actors, including affected communities.” 

193. Deloitte has not tailored its eligibility determination systems to the purpose, nor incorporated human consideration for the adverse and high-risk decisions associated with Medicaid coverage determinations, particularly for the entirely automated procedural terminations, in violation of the Blueprint’s Principle of Human Alternatives, Consideration, and Fallback.  

194. Deloitte does not provide Medicaid enrollees with the logic used to make the eligibility determinations in violation of the UGAI’s Right to Transparency, which includes “access to the factors, the logic, and techniques that produced the outcome.”.  

195. As shown by the repeated experiences across Texas and other states, Deloitte has not ensured the accuracy, reliability, or validity of its eligibility determination software’s eligibility determinations in violation of the UGAI’s Accuracy, Reliability and Validity Obligation. 

196. Deloitte deploys its software without adequate evaluation of its risks and without adequate ongoing monitoring and oversight in violation of the UAGI’s Assessment and Accountability Obligation.  

 E. REQUEST FOR INVESTIGATION AND PRAYER FOR RELIEF 

197. The complaining parties urge the Commission to investigate Deloitte and to find that its use of an inaccurate and unreliable automated system to make Medicaid eligibility determinations constitutes a deceptive and unfair trade practice under Section 5 of the FTC Act.  

198. The complaining parties further urge the Commission to: 
A. Initiate an investigation into the business practices of Deloitte, including specifically investigating whether Deloitte develops, markets, deploys, or maintains TIERS without adequately testing, auditing, or otherwise evaluating TIERS for accuracy and reliability; 

B. Pause Deloitte’s use of TIERS in making Medicaid eligibility determinations until it implements an effective testing, evaluation, and monitoring program that: 

ii. Confirms that TIERS reliably and accurately determines eligibility;  

ii. Will continue to test for and detect errors during deployment, including testing for errors identified in other Deloitte-built Medicaid eligibility determination systems deployed in other states;  

iii. Includes processes to monitor automatic terminations and prevent related wrongful terminations; and  

iv. Provides documentation and periodic, public reporting regarding the monitoring regime and its findings;

C. Require Deloitte to develop a comprehensive harm mitigation strategy that prevents ongoing Medicaid coverage loss once any errors in have been identified, whether through the monitoring regime or otherwise. The harm mitigation strategy should include prohibiting Deloitte from deploying the TIERS system or any related Medicaid eligibility systems, or components thereof, where they impose a significant risk of errors or injury, where the risks cannot be timely eliminated or mitigated, or where there is not enough information to assess the risks of harm. 

D. Require that Deloitte identify individuals previously harmed by errors in TIERS and take corrective action, including addressing harms from loss of Medicaid coverage; 

E. Require that Deloitte make public the design specifications, logic and data sources used to make Medicaid eligibility assessments within TIERS, including making public the specific causes of known errors identified within TIERS;  

F. Require that Deloitte make public the processes, templates, and data sources used by TIERS to produce notices and other communications to Medicaid enrollees, including through the online portal; 

G. Require that Deloitte provide a written notice explaining the precise basis, including the data, data sources, and logic used to reach any Medicaid eligibility determinations or recommendations to any enrollees subject to an adverse action because of the TIERS system; 

H. Require that Deloitte comply with existing public policy frameworks for responsible AI development and use, including the OECD Artificial Intelligence Principles, the Blueprint for an AI Bill of Rights, the Universal Guidelines for AI, Executive Order 14110, and NIST AI RMF; 

I. Require that Deloitte provide such other information or documentation which may be necessary to ensure compliance with the aforementioned monitoring and notice requirements, including but not limited to compliance reports, model cards, and incident reports; and 

J. Provide such other relief as the Commission finds necessary and appropriate.

Respectfully Submitted,

NATIONAL HEALTH LAW PROGRAM

/s/ Sarah Grusin
Senior Attorney
grusin@healthlaw.org

/s/ Elizabeth Edwards
Senior Attorney
edwards@healthlaw.org

/s/ Jane Perkins
Litigation Director
perkins@healthlaw.org

1512 E. Franklin St., Ste. 110
Chapel Hill, NC 27514
Phone: (919) 968-6308

ELECTRONIC PRIVACY INFORMATION CENTER (EPIC)

/s/ John Davisson
Director of Litigation
davisson@epic.org

/s/ Grant Fergusson
Equal Justice Works Fellow
fergusson@epic.org

1519 New Hampshire Ave. NW
Washington, DC 20036
202-483-1140 (tel)
202-483-1248 (fax)

UPTURN, INC.

/s/ Emily Paul
Project Director
emily@upturn.org

/s/ Emma Weil
Senior Policy Analyst
emma@upturn.org

1015 15th St NW, Suite 600
Washington, DC 20005