Asking the AI tool directly about hidden instructions exposed the prompt injection in all 6 security-check iterations, but the authors caution this is not a robust safeguard¶
Claim · All claims
Evidence · 1 study · designr2·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¶
Reading the evidence codes
q is the design tier, i the impact, n the sample, and the kind with its r
the rigour judged within that kind: what each code means.
This claim: no standardized effect is recorded in observations/ for it yet, so no benchmark reading is possible.
q2 i? A basic security check re-using attempt-5 files found and reported the hidden instructions in all 6 iterations (3 docx, 3 pdf), yet the authors state self-checking is not a robust security strategy and may come too late for critical attacks. → Niklas Humble 2026
Evidence¶
Niklas Humble 2026¶
Niklas Humble. (2026). Ethical Implications of Prompt Injection in AI-mediated Grading: An Adversarial Red-Team Evaluation. Journal of Academic Ethics. https://doi.org/10.1007/s10805-026-09758-9
q2 · i? · design · r2
Security check (6 iterations re-using attempt-5 files): Copilot found and reported the hidden instructions "in all instances (Fig. 6)". The authors caution self-checking "is not a robust security strategy" and may not work against all tools or advanced attacks.
"with Copilot finding and reporting the hidden instructions in all instances (Fig. 6). However, it should be noted that prompting an AI-tool to self‑check for successful or attempted prompt injection attacks is not a robust security strategy"
Discussion¶
Related Claims¶
- Basic prompt injection attempts failed or were detected: instruction manipulation alone failed in all 9 iterations, role-playing without concealment was detected in output, and metadata-hidden injections failed in all 9 iterations — related
- Copilot's detection of one advanced injection did not persist: after reporting an embedded instruction once, the same file's injection succeeded in six subsequent iterations without warning — related
- Indirect prompt injections combining instruction manipulation, role-playing, and obfuscation altered AI-assigned grades with attack success rates of 94–100% while remaining undetected — related
- A revised injection combining instruction manipulation and role-playing with an explicit do-not-mention instruction succeeded in all 9 iterations across docx, pdf, and htm formats — related
Page metadata
| Field | Value |
|---|---|
type |
claim |
id |
direct-security-check-exposes-injection |
title |
Asking the AI tool directly about hidden instructions exposed the prompt injection in all 6 security-check iterations, but the authors caution this is not a robust safeguard |
description |
Asking the AI tool directly about hidden instructions exposed the prompt injection in all 6 security-check iterations, but the authors caution this is not a robust safeguard |
status |
review |
evidence_strength |
moderate |
newest_evidence |
2026 |
recheck_by |
2029 |
generated |
by process:wiki-ingest on 2026-10-11 |
trust |
unverified — no one has confirmed this page's content |
sources |
1 — niklas-humble-2026 |
Frontmatter is the page's machine-readable record — see the Schema & Guide.