The State of Play

Education administration and admissions automation has bifurcated into two maturity tracks that define the leading-edge tier. Low-stakes operational automation — transcript processing, financial aid routing, student onboarding, chatbot-based recruitment — is now mature, production-deployed, and ROI-validated across hundreds of institutions. Gartner projects that by 2031, 50%+ of higher education institutions will have fully migrated to cloud SaaS student information systems, signaling ecosystem inflection. Deployment of multi-module platforms (Workday, Ellucian, Dynamics 365) is accelerating: Xavier University's integrated deployment, Carthage College's live onboarding automation, and Aston University's 421-hour-per-month application processing savings document concrete operational gains. AI-assisted insights have shifted from institutional differentiator to baseline expectation: HubSpot integration cases demonstrate 90% faster inquiry response and 42% higher application completion, while AI essay automation at Virginia Tech, Georgia Tech, and Caltech now processes 250k applications per hour with 99.3% accuracy on transcript extraction—reducing per-transcript processing from 5 minutes to 30 seconds. This low-stakes track is deployable territory with clear, measurable ROI and workforce productivity impact.

High-stakes algorithmic screening — allowing systems to influence admission decisions — remains severely constrained. Institutions face documented bias in ML models, reputational risk, legal exposure, and fundamental questions about algorithmic transparency in consequential decisions. Medical school admissions provide cautionary data: NYU Grossman and Zucker School operate ML screening models while applicants must certify non-AI authorship; governance asymmetry creates equity and compliance risk. Most institutions avoid algorithmic screening entirely; those experimenting maintain mandatory human review. This constraint is structural, not temporary, and defines why the practice remains leading-edge rather than advancing toward mainstream. The tier-defining tension is not whether to automate routine tasks (the answer is yes), but how far toward consequential decision-making automation can responsibly extend. For low-stakes operational efficiency, institutions are moving decisively; for high-stakes screening, the line holds. Implementation barriers—particularly extended SIS deployment timelines (2-4 years vs. vendor estimates) and architectural limitations of legacy platforms in supporting emerging operational models—continue to moderate adoption velocity despite growing ROI visibility.

Low-stakes automation has reached inflection-point maturity by early July 2026, with late-June evidence showing workforce-scale deployment. Investigative journalism (GradPilot, June 26) documented AI essay automation at Virginia Tech, Georgia Tech, UNC, and Caltech, with Virginia Tech alone processing 250,000 essays per hour and saving approximately 8,000 staff hours across the 2025-26 admissions cycle. This represents replacement, not capacity addition—Virginia Tech paired one human reader with one AI system after previously employing two humans per essay. SaaS SIS adoption now represents the mainstream trajectory: Gartner projects 50%+ of institutions fully migrated by 2031, and deployment momentum is accelerating. Xavier University's multi-module Workday deployment (HCM, Financial, Student), Carthage College's live student onboarding automation (financial aid, housing, health, parking), and Aston University's documented outcomes (421 hours/month saved on applications) demonstrate production deployments with quantified ROI across institution sizes. Transcript processing automation now delivers 567% productivity gain with 99.3% accuracy, enabling identified detection of 504 previously-invisible qualified applicants at major research universities. EAB's Enroll360 platform now powers 1,200+ partner institutions with 16% average enrollment increase and 17% NTR growth, signaling mature platform adoption at scale.

Market structure shows consolidation continuing around unified platforms. Vendor analysis reports 70% of initial admissions contact now handled by AI agents, with documented enrollment yield increases of 12-18% and cost-per-lead reductions of 20-25% through multi-vendor platform integration (Salesforce, Ellucian, HubSpot). Unified campus portals (Ventura County, Florida Polytechnic) are driving 70% increases in mobile portal usage but raise unresolved privacy and surveillance concerns, creating governance barriers to further adoption. Institutional commitment to responsible AI deployment is systematizing: the National Student Legal Defense Network's governance framework ('10 Dos and Don'ts of AI in College Application Evaluation') signals elevated governance maturity as baseline expectation.

Implementation realities constrain velocity despite ROI visibility. Workday Student deployments routinely extend to 2-4 years (vs. vendor estimates of 18-24 months), revealing integration complexity as the primary cost driver and adoption friction point. WashU's $265M+ total deployment cost across Workday and Student Sunrise illustrates the sunk-cost commitment institutions face, while documented user dissatisfaction suggests implementation challenges persist despite substantial investment. Adoption remains concentrated on low-stakes operational efficiency: approximately 50% of US admissions offices deploy AI for administrative sorting (transcripts, recommendation letters); ~40% use AI detection tools for essay authenticity screening. This distribution signals that institutions view operational automation as deployable, while high-stakes screening remains constrained. High-stakes algorithmic screening continues to face structural barriers: documented bias in ML models with governance asymmetry (applicants required to disclose AI-generated content while institutions operate under voluntary disclosure principles), transparency concerns amplified by the EU AI Act (August 2026 high-risk classification for admissions systems), and institutional unwillingness to cede screening authority to systems whose failure modes carry legal and reputational consequences. Most institutions maintain mandatory human oversight in consequential decisions, a pattern that will likely persist.

Mid-August 2026 evidence confirms accelerating adoption with persistent governance maturity gaps. EduTech survey data shows 68% of major universities now use AI admissions automation (up from 29% in 2024), representing rapid diffusion across the sector. Named deployments continue demonstrating ROI: Georgia Southern's AI agent (integrated with Slate, Banner, PeopleSoft) handled 300k+ student inquiries in two months with 2% enrollment growth and $2.4M projected additional revenue; Columbus State achieved 75% reduction in wait times and 85% first-contact resolution. EDMO platform reports processing 3M+ student documents with $10M+ aggregate institutional cost savings, indicating scaled adoption across six named institutions and four major CRM ecosystems (Salesforce, Slate, HubSpot, Zoho). However, governance infrastructure has not kept pace with deployment velocity: a Student Defense public records survey found zero policies governing AI use in undergraduate admissions at 24 major public universities, and zero formal AI training for admissions staff. Compliance frameworks remain inadequate: independent technical review identifies critical gaps in FERPA audit logging and data residency controls across leading enrollment platforms, with assessment noting that "deployment pace outstrips compliance frameworks." A cautionary counterweight emerges from UNAM's production failure: the university's 2026 AI-proctored entrance exam for 160k applicants experienced implausible score inflation (120-point exam: 3.5% scoring 100+ in 2021-25 vs. 16.3% in 2026), statistical analysis suggested ~50% cheating rates despite AI oversight, and the university mandated ~58k retakes—demonstrating that automation of high-stakes testing creates new failure modes and reputational risk. The evidence picture remains bifurcated: low-stakes operational automation is proven, scaling, and ROI-validated; governance infrastructure and institutional readiness lag significantly behind deployment adoption; high-stakes algorithmic decision-making remains constrained by well-documented fairness concerns and regulatory scrutiny.

• 2019: Early adoption of chatbot-based recruitment and retention platforms (AdmitHub 50+ customers) and first-generation financial aid verification automation (Campus Management 50% processing time reduction at UW-Madison). Parallel emergence of concerns about algorithmic bias in admissions decisions as a regulatory and equity risk.
• 2020: Continued vendor expansion (AdmitHub 90+ institutions, $7.5M Series A from Google/Salesforce) and deployment of chatbot-based student support during COVID-19 pandemic (Common App partnership). Critical setback: University of Texas at Austin discontinued its GRADE algorithmic screening system after 7 years despite documented 71% efficiency gains, citing bias concerns. Signals market divide between viable low-risk administrative automation and constrained high-stakes screening use cases.
• 2021: Consolidation of market into two adoption tracks. Low-risk chatbot and CRM-based admissions automation scaled: AdmitHub $14M Series B, 3M+ student interactions, 30% summer melt reduction at Georgia State; Indian edtech platform achieved 30% lead-to-enrollment gains via LeadSquared; Australian deployment via Salesforce EDA. Infrastructure reliability concerns emerged: UC/Cal State admissions portals crashed during peak application deadlines. Regulatory scrutiny intensified: New America documented how predictive analytics perpetuate racial inequities; UK Government AI Barometer identified bias as primary adoption barrier. High-stakes algorithmic screening remained constrained by equity concerns.
• 2022-H1: Continued low-risk automation scaling (AdmitHub/Mainstay 100+ institutions); emergence of alternative admissions innovation (Concourse portfolio-based system, 658+ pilot participants, $18.1M scholarships). Academic research renewed focus on algorithmic bias: MIT documented fairness gaps up to 21% in ML explanation methods for admissions; Brookings analysis showed enrollment algorithms reduce aid for low-income students. Institutional response: NSF/Amazon funded $1M project for fair algorithms in graduate admissions. Market divide between low-stakes automation (scaling) and high-stakes screening (constrained by fairness requirements) solidified, with investment shifting toward equity-focused solutions.
• 2022-H2: Institutional adoption of chatbot automation continued (Texas A&M System deployment of AdmitHub), confirming sustained multi-institution rollout through year-end. However, adoption barriers persisted: survey research found majority of HEIs not ready for AI chatbot deployment, highlighting persistent organizational, technical, and trust gaps in the market despite vendor traction.
• 2023-H1: Low-risk automation platforms demonstrated sustained deployment scale (Family Legacy Foundation AdmitHub chatbot for 1,000+ students annually, Antioch University text automation achieving 62% lead-to-enrollment gains). Institutional investment in enrollment data infrastructure accelerated (Northeastern University multi-year modernization via Snowflake and dbt). Research on fairness limitations intensified: UT Austin study found ML models for student-success prediction systematically less accurate for racially minoritized students with ineffective mitigation; Brookings documented enrollment algorithms reduce scholarships and perpetuate discrimination. Counterbalancing innovation: Cornell research demonstrated ML admissions models can outperform SAT-based screening while maintaining demographic parity. Persistent market divide: low-stakes automation scaling with clear ROI; high-stakes algorithmic screening adoption constrained by fairness and equity barriers.
• 2024-Q1: Generative AI impact on admissions became visible: institutional caution toward algorithmic screening intensified (SMU explicitly discontinued personal statements due to ChatGPT but rejected AI-based screening), while low-risk automation continued evolving (UC Irvine integrated ChatGPT-powered chatbots with 96% accuracy, Columbia College of Missouri deployed AI funnel management improving staff morale). Landscape survey data showed early-stage adoption: Kaplan survey found only 14% of admissions officers use AI in their work, 85% lack GenAI policies, 9% use detection software—indicating institutional uncertainty despite growing vendor capabilities. Peer research confirmed persistent barriers: technical, ethical, and resource challenges limit scaling beyond specialized low-risk use cases.
• 2024-Q2: No-code and specialized automation tools entered production deployments: AppSheet case study showed significant processing time reductions in university admissions workflows; Freedom OCR platform gained adoption for transcript extraction and SIS integration. Institutional adoption of bot-based admissions automation continued scaling per UCAS conference case studies. High-stakes algorithmic screening remained cautious: University of Miami piloted AI for application volume handling (50,000+ applications), while IE University deployed AI for major selection—but institutional hesitation persisted. Critical assessment documented negative impacts: 2024 admissions cycle revealed increased opacity in AI-driven selection processes, FAFSA processing failures, and 40% decline in financial aid completion. Market division reinforced: low-stakes automation demonstrated value; high-stakes screening constrained by equity and transparency concerns.
• 2024-Q3: Production deployment of low-risk automation expanded: Georgia Tech screened 60,000+ applications annually with expanded admissions team, and CollegeVine onboarded 50 institutional partners for agentic AI recruiting. Peer-reviewed research from AERA demonstrated that algorithmic bias in student-success prediction models persists despite mitigation efforts, with disparate accuracy across racialized groups. Adoption survey data (Martin Center, James G. Martin Center) indicated 50% of US admissions offices use some form of AI, but critical assessment emphasized transparency gaps and regulatory risks. Market division solidified: low-stakes automation (lead management, chatbots, transcript processing) deployed at scale with documented ROI; high-stakes algorithmic screening remained constrained by well-documented fairness gaps and institutional caution. Vendor consolidation continued around recruitment and support tools rather than algorithmic screening.
• 2024-Q4: SIS vendor ecosystem reached inflection point: 28 institutions deployed Workday Student in Q4, adding to $15.3B-to-$32B market growth forecast. Adoption survey (Liaison, Oct 2024) confirmed institutional reliance on conversational AI (73%) while strategic adoption of predictive (41%) and prescriptive (17%) AI remained limited. Expert panel discussion (Inside Higher Ed, Dec 2024) highlighted AI potential for credit transfer automation, tuition discounting, and student success—but implementation barriers persisted. Infrastructure reliability risks emerged during peak admissions processing (Ellucian outage, Nov 2024). Market remained bifurcated: low-stakes automation (chatbots, lead routing, transcript processing via Freedom OCR) continued scaling with documented ROI; high-stakes algorithmic screening constrained by persistent fairness research and institutional caution. No major policy shifts or regulatory changes in window, but transparency and liability concerns remained top institutional barriers.
• 2025-Q1: Low-risk automation continued demonstrating ROI in early 2025: six institutions reported concrete gains (Empire State 25% engagement increase and 4% retention, Bakersfield College $2M+ savings, Long Beach City College 10x ROI). Mainstay (formerly AdmitHub) remained in production use with expanded feature set. Institutional sentiment shifted measurably: Acuity Insights survey of 160 admissions leaders found 51% believe AI will transform applicant evaluation, reflecting confidence growth in the market. However, fairness research persisted as a critical counterweight: new peer-reviewed study analyzed real university admissions data (11,600 students) confirming ML models exhibit demographic bias—white and non-first-gen students incorrectly admitted at higher rates despite test-optional policy. Practitioner discourse emphasized bias risks and mitigation strategies. Market outlook remained stable: low-stakes chatbot and CRM-based automation gained confidence and ROI evidence; high-stakes algorithmic screening continued constrained by documented fairness gaps and institutional caution.
• 2025-Q2: SIS vendor consolidation accelerated with Workday Student deployments at LSU (Spring 2025) and Colby College (April 2025), though implementation challenges emerged around UX design and feature completeness despite backend efficiency gains. Podcast-driven market landscape analysis (April 2025) provided concrete deployment metrics: NYU 120k applications, UTexas +24% growth, UWash +57% growth. Peer-reviewed systematic review (June 2025) on AI in applicant screening analyzed current tools and bias concerns. Practitioner critical assessment (May-June 2025) reinforced focus on fairness audits, explainable AI, and hybrid human-review models. Market divide persisted: low-stakes automation (chatbots, transcript processing) continued scaling with proven ROI; high-stakes algorithmic screening remained constrained by peer-reviewed bias research and institutional fairness concerns. Regulatory environment and Supreme Court ruling implications (2023) kept institutional risk calculus cautious around autonomous AI screening systems.
• 2025-Q3: Policy-level automation innovation demonstrated impact: UT Austin pilot combining automatic admissions eligibility with proactive financial aid guarantees nearly doubled enrollment for eligible low-income students (23% to 43%), validating automated policy-implementation systems. Global market expansion accelerated: CSM Technologies deployed Student Academic Management System (SAMS) in Indian states (Odisha, Bihar) for centralized admission automation; Uttar Pradesh government launched Samarth portal for state-level university admissions consolidation. Market analysis showed USD 1.58B admission management system market (2024) projected to reach USD 4.55B (2034) at 9.8% CAGR, indicating sustained institutional and vendor investment in automation infrastructure. Implementation barriers persisted: Workday deployments continued facing UX and feature completeness challenges despite operational efficiency gains. Fairness research and practitioner guidance remained consistent through Q3: focus on audits, explainable AI, and hybrid models to mitigate bias in higher-stakes screening systems.
• 2025-Q4: Low-risk automation continued demonstrating concrete ROI with peer-reviewed research validation: October 2025 systematic review synthesized AI in academic applicant screening across undergraduate to medical residency programs, confirming widespread adoption. Institutional deployments at scale reached production maturity: Georgia State's AdmitHub chatbot processed 99% of 50,000 student messages with 94% satisfaction; Salesforce EDA implementations cut admissions response time 26%; Southeast Missouri State saved 182 staff hours monthly via AI chatbots. Enrollment software vendor ecosystem solidified: industry analysis showed 567% productivity gains and 99.3% accuracy with AI-powered transcript processing, indicating tool maturity. Market momentum accelerated: vendor comparative reviews identified 10% enrollment uplift from AI screening tools with documented 250+ annual man-days savings. Persistent implementation challenges offset efficiency wins: Workday Student rollouts continued facing UX complexity and feature gaps. High-stakes algorithmic screening remained constrained by sustained peer-reviewed fairness research and institutional caution; practitioner guidance (May-June 2025) emphasized explainable AI, fairness audits, and hybrid human-AI models as essential safeguards. Market divide solidified by year-end: low-risk chatbot and transcript automation continued scaling with proven ROI; high-stakes algorithmic screening remained cautious due to well-documented bias risks and institutional liability concerns.
• 2026-Jan: January deployments at Caltech (VIVA AI interview tool) and Virginia Tech (AI essay evaluation) demonstrated continued adoption of AI evaluation tooling alongside human review. International student survey (1,600+ respondents) found 17% use AI for university search with high satisfaction (96% met/exceeded traditional sources), signaling student-side adoption shift. Peer-reviewed research (AAAI 2026) documented technical limitations: ML admissions models show degraded performance when applicant pool composition shifts, confirming reliability constraints in algorithmic screening. Practitioner guidance emphasized ongoing challenges in AI-generated writing detection for academic integrity in graduate admissions.
• 2026-Feb: Research on AI-generated essay writing in applications (81,663 applicants, 2020-2024) showed LLM adoption accelerating in 2024 with concerning equity patterns: lower SES students used LLMs at higher rates but faced stronger admission probability declines. Augsburg University demonstrated positive alternative: automatic admission program reduced application friction and improved student confidence without algorithmic screening. Private school survey indicated 87% portal adoption and 77.8% online scheduling, confirming partial automation of routine intake tasks while preserving human touchpoints. Admissions system modernization continues with reported 25% cycle acceleration and 35% manual task reduction via CRM platforms. Security and reliability concerns persisted: guidance documents cited institutional errors (mistaken acceptances) and data breaches (Columbia). Market momentum remained stable: low-stakes automation demonstrating continued ROI and scaling; high-stakes algorithmic screening constrained by both fairness research and emerging AI-generated writing authenticity challenges.
• 2026-Mar: Market surveys confirm AI workflow automation and CRM integration as baseline expectations across leading enrollment platforms (Slate 9.1/10, Element451 8.8/10 in 2026 vendor rankings). Institutional adoption sentiment has shifted measurably: Ellucian's survey of 779 professionals across 300+ institutions shows 66% institutional AI adoption (up from 49%), with 51% citing Marketing/Admissions/Enrollment as top benefit area; Comm100 report documents 44% AI deployment in admissions with 6-10 staff hours saved weekly and 1-6 month payback. Response-speed automation confirmed as a material conversion driver with research-backed evidence linking faster lead routing to enrollment uplift. Low-stakes automation continues scaling with strong ROI evidence; high-stakes algorithmic screening remains constrained by fairness concerns and institutional caution.
• 2026-Apr: SIS vendor ecosystem maturity confirmed at new scale: Workday Student now manages 5.8M student records across 200+ institutions globally; Ellucian Student launched as a unified AI-native platform integrating SIS, HCM, and Finance (April 2026 GA), with a record 26 SaaS go-lives in Q1 2026 (including Aurora, Mohamed bin Zayed AI University, and Norwich). Gartner Magic Quadrant 2026 recognized both Workday and Ellucian as leaders, with Workday's Pensacola State deployment achieving 43% enrollment growth. AI evaluation tools matured operationally: Virginia Tech's essay reader processes 250k applications in <1 hour, Caltech's VIVA bot tests intellectual ownership, California Community Colleges deploy fraud detection—all with human oversight preserved. Practitioner landscape research identified growing equity concerns: EAB survey of graduate enrollment managers shows 87% have tried AI but only 23% of institutions formally use it, signaling organizational adoption lag; Foundry10's first full-cycle study finds students from families earning $75k–$100k adopt AI in applications at 150% higher rate than lower-SES peers, raising fairness questions. Governance advancement: ACM FAccT 2025 introduces hyperFA*IR algorithm addressing finite-pool selection bias in university admissions systems. EAB enrollment leaders survey shows 30% prioritize "Actually-Existing Admissions AI," seeking hype-busting examples and real ROI validation. Market remains bifurcated: low-stakes automation (chatbots, lead routing, transcript processing) proven and scaling; high-stakes screening constrained by fairness research, AI-generated essay authenticity questions, and regulatory emergence (EU AI Act August 2026 classification as high-risk).
• 2026-May: SIS migration momentum is at peak scale: Swarthmore (October 2026 Phase 1), Carthage College (fall 2026 onboarding automation), MSU Denver (phased 2025–2027), and the University System of Georgia (July 2028 go-live across 25 institutions — the largest multi-institutional Workday commitment yet). Central Michigan University's Slate CRM integration achieved first-in-Michigan early financial aid delivery and automated scholarship matching. Market structure research explains why automation pressure is intensifying: AI-assisted college search adoption doubled to 46% in one year (tracking toward 80% by 2028), selective-institution yield fell 14 points since 2019, and cost per inquiry is rising 30% — all driving institutional investment in enrollment automation. New institutional adoption data shows 74% of US institutions now have production AI deployments touching students (up from 28% in 2024), with admissions chatbots and applicant Q&A among the confirmed use cases. Equity concerns sharpened: a synthesis of four empirical studies (Cornell, Stanford, Foundry10) found lower-SES students face 31% larger admission gaps and 1.85x steeper penalties per unit of AI essay use, despite adopting AI at similar rates. EAB analysis of 7M+ journeys confirms 53% of prospective students narrow their college list before any CRM contact, challenging the assumption that yield-stage automation operates at the critical decision point.
• 2026-Jun: Operational automation ROI is now well-documented at production scale: Aston University and TEDI-London report 421 and 800 hours/month saved respectively via Dynamics 365 admissions automation, and EAB's Enroll360 (1,200+ partner institutions) delivers 16% average enrollment increase. Gartner's prediction that 50%+ of institutions will fully migrate to cloud SaaS SIS by 2031 gained credibility as vendor consolidation continued — Ellucian named Leader for the second consecutive year with agentic workflows on its roadmap. The bifurcation between low-stakes operational automation (deployable, ROI-confirmed) and high-stakes algorithmic screening (constrained by bias, transparency, and the National Student Legal Defense Network's published governance framework) remains the defining structural feature of the practice.
• 2026-Jul: Workforce-scale automation at the low-stakes end is now documented fact: investigative reporting confirmed Virginia Tech processes 250,000 essays per hour saving approximately 8,000 staff hours, with peer institutions (UNC, Georgia Tech, Caltech) following; AI agents handle 70% of initial admissions contact across leading platforms, delivering 12–18% enrollment yield increases and 20–25% cost-per-lead reductions. Texas Tech and Stony Brook's AI transcript processing achieved 567% productivity gains with 99.3% accuracy, surfacing 504 previously-invisible qualified applicants. Against these operational gains, structural governance asymmetry persists: NYU Grossman and Zucker operate ML screening while requiring applicants to certify non-AI authorship, and unified-portal deployments raising privacy and surveillance concerns signal that equity and compliance risks remain structural rather than transitional barriers. Mid-July evidence confirms continued vendor consolidation: Workday Student now manages 5.8M student records across 200+ institutions and was named Gartner Magic Quadrant Leader for a second consecutive year, while the global SIS market is projected to grow from $20.72B (2026) to $67.12B (2034); enrollment-chatbot case studies (Instadesk's Lone Star College deployment at 96% accuracy, a $29.7M LTV lift at a national healthcare university, SWPS University's 200+ automated communication workflows) reinforce low-stakes ROI at scale. A legal-liability framework sharpened around high-stakes screening: analysis cites Colorado SB 26-189 and precedents (Mobley v. Workday, Newby v. Adelphi) establishing institutional responsibility for algorithmic admissions decisions. Late-July vendor-ecosystem evidence reinforced SIS consolidation momentum: Huron was named a Strong Performer in Forrester's Q2 2026 Workday Services Wave, and Augsburg University completed a full Workday cloud-ERP migration. Counter-signals on integration quality emerged in parallel: Ellucian's Banner API was rated "Grade F" by developer reviewers, while a coaching-center CRM case study documented 6x enrollment growth—underscoring continued bifurcation between mature vendor infrastructure and uneven implementation quality.
• 2026-Aug: Category-wide adoption survey data confirms rapid maturation: 68% of major universities now use AI admissions automation (up from 29% in 2024), with named deployments (Manchester 48% manual-review reduction, Stanford 93% compliance detection); EDMO reports 3M+ student documents processed and $10M+ savings across integrations with four major CRMs (Salesforce, Slate, HubSpot, Zoho) at NYU, UPenn, and other national institutions, and Druid AI documents measured ROI at Georgia Southern (2% enrollment growth, $2.4M projected revenue) and Columbus State (75% wait-time reduction). Governance lag remains the defining counter-signal: a Student Defense public-records survey found 0 of 24 public universities have written AI admissions policies or provide staff AI training, and independent technical assessment concludes deployment pace outstrips compliance frameworks, citing FERPA and data-residency gaps across leading enrollment platforms. A high-stakes failure case reinforces the risk: UNAM's AI-supervised remote exam for 160k applicants required ~58,000 retakes after statistical analysis suggested roughly 50% cheating despite AI oversight—a concrete cautionary data point for automated admissions administration at scale.