{"resourceId":"athletic-cfb-recruiting-ai-film-triage-2026","versions":[{"version":"external-8609ab6071f040682ddb0b3d83687f5721ae6f21ba31762c45a8dfd6f8ea5c54","resource":{"id":"athletic-cfb-recruiting-ai-film-triage-2026","title":"Power and G6 staff describe AI film triage as a filter amid deepfake-highlight risk","organization":"The Athletic (New York Times)","sector":"College athletics / football recruiting operations","geography":"United States (Power 4 and Group of 6 programs; anonymous sourcing)","publishedAt":"May 1, 2026","publicationDate":"2026-05-01","eventDate":null,"sourceName":"The Athletic","sourceLabel":"Independent reporting — college football recruiting AI","sourceUrl":"https://www.nytimes.com/athletic/7244027/2026/05/01/college-football-recruiting-ai-power-4-humans/","evidenceClass":"independent-reporting","outcomeClass":"mixed","topics":["developers-agents","knowledge-work","operating-model","data-security"],"finding":"Antonio Morales reports Power 4 and Group of 6 football personnel using or considering AI to grade highlight film and shrink evaluation lists, while other staff doubt AI can capture movement qualities or character and warn that AI-generated fake highlights could disrupt evaluation. Quoted practitioners insist AI should act as a filter, not a replacement for real or live evaluation, and note creative use for quick recruiting graphics.","sledRelevance":"Why now: 2026-09-15 edition left recruiting/compliance AI outcomes as an open gap after covering coaching agents, fatigue ML, and wearable consent. This still-unarchived May 2026 Athletic investigation supplies current collegiate football recruiting practice evidence—efficiency claims, human overrides, and synthetic-film risk—amid ongoing portal/NIL roster pressure.","evidence":"Anonymous P4 personnel describe an alum-built agent trained on hundreds of hours of film using house grades; a G6 GM describes filtering thousands of prospects toward a few hundred; skeptics cite culture/makeup and movement qualities; a February 2026 social post of AI-fabricated highlights is discussed as an emerging threat staff have not yet confidently detected.","architectureImplications":"Film-ingest scoring agents need labeled grade rubrics, human accept gates, and synthetic-media detection—not autonomous signing recommendations.","governanceImplications":"Athletic departments should document AI-as-filter policy, override ownership when models contradict coaches, and brief compliance/comms on deepfake recruiting media.","securityPrivacyImplications":"Prospect film and PII in triage pipelines require access control; inbound media authenticity checks become part of recruiting security.","caveats":"Anonymous sourcing; no measured signing-class outcomes; efficiency claims are practitioner-reported; deepfake prevalence unquantified.","streamIds":["college-athletics"],"roles":{"sales":"Sales framing: Power and Group-of-6 football programs are already experimenting with AI film triage and portal list-building under roster churn, but anonymous Athletic reporting shows staff split between efficiency gains and deep skepticism about quantifying movement, body type, and character. Position any offering as a supervised filter that shrinks the watch list—not as an autonomous evaluator—and price proof against time saved per scout plus false-negative risk on prospects they later wish they had seen. Do not sell culture-fit or five-star persuasion as an AI deliverable; relationships remain the bottleneck the piece emphasizes.","engineering":"Engineering framing: Feasible architectures resemble highlight-ingest agents scored against a program’s existing grading rubric, with human review gates before a prospect enters deeper evaluation. Integration points include film libraries, measurables (40, broad jump), and competition-level production tables—not a black-box “grade the kid” endpoint. Prerequisites are labeled training film, documented grade definitions, and QC loops when the model drifts. Constraints include adversarial AI-generated fake highlights, weak encoding of lateral mobility and stiffness, and staff refusal to trust character columns. Proof of value is precision/recall on held-out graded film plus scout override rates, not vendor demos.","delivery":"Delivery framing: Stand up a shadowed film-triage pilot with personnel directors owning accept/reject of AI shortlists, creative staff optionally using AI for lightweight recruiting graphics, and compliance/communications briefed on deepfake-highlight risk. Dependencies: alum or vendor model access, film storage permissions, and written rules that AI never replaces live evaluation. Skills adoption centers on teaching staff to specify filters rather than seek “all the answers.” Governance checkpoints include quarterly false-highlight drills and documented overrides when AI contradicts a coach. Acceptance criteria: measurable reduction in unwatched low-value clips without missed priority targets agreed by GMs."},"roleDetails":{"sales":{"customerProblem":"Football recruiting staffs face thousands of highlight submissions and portal targets while needing faster shortlists without surrendering evaluation quality or culture assessment.","stakeholders":["Personnel directors / GMs","Position coaches","Creative/recruiting communications","Compliance (deepfake film risk)"],"discoveryQuestions":["What grade rubric would train a film-triage agent?","How many highlights are watched vs ignored today?","Who overrides an AI shortlist?","How do you detect manipulated or AI-generated clips?"],"valueHypothesis":"A program-tuned filter can shrink watch lists (e.g., toward top hundreds) if humans still own final grades and live evaluation—matching how quoted G6/P4 staff describe intended use.","potentialEngagement":"Shadowed film-triage pilot on one position group with precision/recall against existing staff grades.","unsupportedClaims":["AI replaces live evaluation or character assessment","Guaranteed better signing classes","Deepfake-proof film intake"]},"engineering":{"fit":"Supervised or agentic highlight scoring aligned to house grades, plus optional generative graphics; not end-to-end autonomous recruiting.","architectureIntegration":"Ingest recruit film and measurables; output ranked shortlists with feature rationales into existing CRM/board tools; keep humans in the accept path.","prerequisites":["Labeled historical grades","Film storage access","Defined filter features","QC process for model drift"],"constraints":["Hard-to-encode movement qualities","Culture/makeup gaps","Adversarial AI fake highlights","Uneven G6 vs P4 tooling budgets"],"security":"Control access to prospect PII and film; log who accepted AI grades; monitor inbound media for synthetic artifacts.","proofOfValue":["Agreement with staff grades on held-out film","Time per evaluated prospect","Missed-target rate after filtering","Detected synthetic-clip incidents"]},"delivery":{"work":"Configure filters, train staff on shortlist review, add deepfake-awareness checks, and measure override patterns for one recruiting cycle.","dependencies":["Personnel leadership buy-in","Creative team guidelines","Legal/compliance review of outbound AI graphics"],"ownership":"Recruiting/personnel leadership owns decisions; analytics supports the filter.","skillsAdoption":"Teach specifying filters and challenging model output rather than accepting AI as final evaluator.","governanceCheckpoints":["Written AI-as-filter policy","Quarterly deepfake drill","Documented coach overrides"],"acceptanceCriteria":["Agreed reduction in low-value watches","No increase in missed priority targets","Synthetic-clip response runbook tested"],"risks":["False confidence in AI grades","Recruit film fabrication","Staff cultural rejection","Leak of proprietary grade models"]}},"retrievedAt":"2026-09-17T03:06:10Z","enrichedAt":"2026-09-17T03:06:10Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"G6 vs P4 tooling budgets differ; creative staff may gain quick-graphic capacity while personnel roles need new QC skills for AI shortlists and fake film.","procurementImplications":"Buy or build against house grade data and require human-in-the-loop clauses; treat generative graphics tools separately from evaluation agents.","operatingModelImplications":"Redistributes early screening work toward analysts/agents while preserving coach relationship work as the scarce resource the piece emphasizes.","sourceVerification":{"openedUrl":"https://www.nytimes.com/athletic/7244027/2026/05/01/college-football-recruiting-ai-power-4-humans/","referenceExcerpt":"It needs to be used as a filter.","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}