{"resourceId":"vt-swag-writing-ai-accountability-chi-20261008","versions":[{"version":"external-9dfeaf4ea0d41638eb29e9244b87ad348df5a6c9cdd8ac9cde8311bd59b1c0ac","resource":{"id":"vt-swag-writing-ai-accountability-chi-20261008","title":"Virginia Tech: nearly half of AI writing interactions asked ChatGPT to write the essay; recording interactions cut outsourcing even when anonymized","organization":"Virginia Tech Department of Computer Science (Sang Won Lee, Andrew Jelson, Dan Dunlap and colleagues; reported by Virginia Tech News)","sector":"Higher education student success / generative AI writing and academic integrity","geography":"United States (Virginia Tech classroom studies)","publishedAt":"Virginia Tech News article dated 8 October 2026 summarizing a CHI 2026 proceedings paper and a separate classroom SWAG deployment","publicationDate":"2026-10-08","eventDate":null,"sourceName":"Virginia Tech News","sourceLabel":"University news summary of a peer-reviewed CHI 2026 study (n=77) and a subsequent classroom deployment of the SWAG accountability platform","sourceUrl":"https://news.vt.edu/articles/2026/09/eng-cs-writing-on-the-wall-AI-education.html","evidenceClass":"independent-reporting","outcomeClass":"cautionary","topics":["knowledge-work","governance-procurement","operating-model"],"finding":"A Virginia Tech–led team built an online writing environment that logged every ChatGPT question, response and copy-paste while 77 students wrote an assigned essay with unrestricted AI access. Six students used no ChatGPT; among the rest, nearly half of all AI interactions asked the tool to write the essay, with others brainstorming, proofreading or 'vibe writing.' Students who used AI mainly for planning or revising reported ownership similar to non-users; only those who asked ChatGPT to generate the entire essay reported significantly lower ownership. In a later classroom deployment of Student Writing with Accountable Generative AI (SWAG), students knew instructors could see AI interaction records; requests for the model to write an essay on their behalf dropped significantly, and the drop persisted when submissions were anonymized. SWAG can apply instructor-set guardrails (for example declining whole-essay requests). The team is integrating with LMS platforms for assignment-specific allowed assistance and transparency reports.","sledRelevance":"Published today, it gives rare instrumented evidence of how undergraduates actually use unrestricted ChatGPT for writing and shows that making AI use visible to instructors changes behavior without relying only on detectors. It speaks to durable writing skill and integrity rather than engagement alone.","evidence":"CHI 2026 proceedings study with 77 students and interaction logging; university news summary of a separate SWAG classroom deployment with observed drop in whole-essay requests under visibility, including anonymized condition; quotes from Lee, Jelson and Dunlap; DOI pointer 10.1145/3772318.3791056.","architectureImplications":"Writing environment that logs prompts, responses and paste events; SWAG adds instructor-visible transcripts, assignment guardrails and planned LMS integration with transparency reports.","governanceImplications":"Interpretation: shift from post-hoc detection to assignment-level allowed-use rules plus visible logs students know about; disclose recording before use.","securityPrivacyImplications":"Logging every keystroke-level AI interaction creates sensitive educational records; production use needs notice, retention limits and access rules for instructors and admins.","caveats":"News summary rather than the full CHI methods tables; SWAG classroom results are described without sample size or exact percentages in the article; single-institution context; ownership findings are self-reported.","streamIds":["student-success"],"roles":{"sales":"The framing point is visibility over detectors. Virginia Tech logged unrestricted ChatGPT use while students wrote essays and found nearly half of AI interactions asked the tool to write the essay; when SWAG made logs visible to instructors, whole-essay requests dropped even if work was anonymized. Use it with writing programs weighing accountability platforms. Bound it: news summary of one campus, self-reported ownership, and limited SWAG sample detail in the article.","engineering":"For engineering, the framing point is assignment-scoped logging and guardrails. Capture prompts, responses and paste events, let instructors decline whole-essay requests per assignment, and plan LMS-linked transparency reports that sit beside the submitted essay. Treat interaction logs as sensitive education records with clear student notice, short retention windows and role-based instructor access before any production rollout on campus writing courses.","delivery":"Delivery's framing point is changing the task norm. Pilot SWAG-like logging in a few writing courses, tell students up front that AI interactions are reviewable, set allowed-assistance rules per assignment, and track whole-essay outsourcing rates before and after. Do not treat a drop in detector flags alone as success if logs still show answer outsourcing."},"roleDetails":{"sales":{"customerProblem":"Faculty cannot see how students use consumer AI for writing, so detectors miss process and students outsource essays while feeling ownership.","stakeholders":["Writing program or composition director","Provost or associate provost for undergraduate education","Academic integrity office","LMS / academic technology","Faculty senate"],"discoveryQuestions":["What share of writing assignments allow AI, and how is that stated?","Do you have any process logs, or only similarity/AI detectors?","Would faculty review interaction summaries if students were notified?"],"valueHypothesis":"Assignment-scoped visibility and guardrails may reduce whole-essay outsourcing without a ban; transfer beyond Virginia Tech is unproven.","potentialEngagement":"A writing-program pilot of accountable AI logging with before/after outsourcing measures on a few courses.","unsupportedClaims":["That logging eliminates cheating","That ownership findings generalize to all disciplines","That SWAG raises learning outcomes"]},"engineering":{"fit":"Institutions with LMS writing courses willing to notify students and scope AI assistance by assignment.","architectureIntegration":"Writing or LMS-integrated AI assistant with prompt/response/paste logging, instructor-configured refusal rules and exportable transparency reports.","prerequisites":["Student notice and consent language","Role-based instructor access","Assignment policy configuration UI"],"constraints":["Privacy and retention review","Students may switch to unsanctioned tools","Single-campus evidence in the source"],"security":"Minimize retention of full transcripts, restrict who can open logs, and keep logs out of model training.","proofOfValue":["Rate of whole-essay generation requests before vs after visibility","Share of assignments with explicit AI rules","Student comprehension of logging in a brief survey"]},"delivery":{"work":"Run a multi-section writing pilot with notified logging, assignment guardrails and weekly review of outsourcing metrics.","dependencies":["Writing faculty champions","LMS integration or stand-alone SWAG-like tool","Privacy counsel sign-off"],"ownership":"Writing program owns policy; academic technology owns the platform; instructors review course-level reports.","skillsAdoption":"Short faculty workshops on reading transparency reports and setting allowed-assistance levels.","governanceCheckpoints":["Privacy review of logging","Mid-pilot outsourcing metric check","Decision to expand or stop"],"acceptanceCriteria":["Students notified before first use","Whole-essay request rate measured and reduced vs baseline","Assignment AI rules published for pilot courses"],"risks":["Migration to unmonitored consumer tools","Faculty underusing reports","Student trust loss if purpose of logs is unclear"]}},"retrievedAt":"2026-10-09T03:30:01Z","enrichedAt":"2026-10-09T03:33:56Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Visibility and guardrails may help students practice accountable AI use employers expect, but surveillance concerns need clear notice and purpose limits.","procurementImplications":"Prefer writing-support tools that support assignment-scoped policies and instructor-visible logs over detector-only products.","operatingModelImplications":"Set allowed AI assistance per assignment, tell students logs are reviewable, and measure whole-essay outsourcing rather than binary AI detection.","sourceVerification":{"openedUrl":"https://news.vt.edu/articles/2026/09/eng-cs-writing-on-the-wall-AI-education.html","referenceExcerpt":"In nearly half of all AI interactions, students asked the tool to write the essay for them.","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}