AI agent activity in an LMS is indistinguishable from normal student activity, defeating detection and proctoring¶

Claim · All claims
Evidence · 1 study · theoretical r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? The company behind Blackboard acknowledged that AI agents look identical to normal student activity, rendering them undetectable to the platforms they inhabit. → Aras Bozkurt 2026

Evidence¶

Aras Bozkurt 2026¶

Aras Bozkurt, Helen Crompton, & Caroline Fell Kurban. (2026). The Devil is in the Det[ai]ls: AI Agents, Ghost Students, and the Crisis of Verified Presence in an Agentic AI World. Open Praxis, 18(1), 1–12. https://doi.org/10.55982/openpraxis.18.1.1145

q2 · i? · theoretical · r2

The editorial cites an Anthology statement from late 2025 that AI agents look identical to normal student activity, supporting its argument that proctoring cannot distinguish agent from human. No effect size applies.

"As Anthology, the company behind Blackboard, acknowledged in late 2025, AI agents "look identical to normal student activity" (Burrett & Gore, 2025), rendering them undetectable to the very platforms they inhabit"

Discussion¶

Related Claims¶

• AI agents can now navigate and complete tasks within learning management systems — a broader claim this one bears on
• In a human Turing test, participants could not reliably distinguish BEAGLE traces from real student data (52.8% accuracy; d′ = 0.15 within the ±0.3 equivalence bound, pTOST = 0.038) — related
• The test platform detects rapid guessing in real time and attempts to re-engage disengaged students to protect score validity — related
• After humanization with Undetectable AI, fewer than 4% of AI-labeled rewrites remain detectable (FNR > 96%) for Pangram and GPTZero — related
• Students cannot reliably distinguish AI-generated slides from human-created slides — related