Standards of
Governance & Integrity
RSIT is not just a school. It is a credentialing authority for practitioners who will govern high-risk AI systems in regulated environments. Our academic standards do not mirror enterprise compliance — they define it.
The Human-in-the-Loop Mandate
In the era of Generative AI, academic integrity does not mean abstaining from technology. It means mastering control over it. The old concept of cheating — a student copying from another student — is obsolete. The new risk is far more dangerous: a practitioner who cannot distinguish their own reasoning from an AI's hallucination, and who is about to govern a critical system.
At RSIT, we enforce a strict Human-in-the-Loop (HITL) academic policy. You are permitted — and often encouraged — to use AI agents as force multipliers for research, synthesis, and code generation. However, you must function as the Senior Auditor of every output before it enters a submission.
This is not an honor system. It is a competency verification system. The discipline you build here — interrogating, auditing, and overriding AI — is precisely what your future employers and regulators will require you to demonstrate under the EU AI Act, ISO 42001, and enterprise AI governance frameworks.
You may delegate tasks to AI. You may never delegate responsibility. If your AI hallucinates a citation, generates vulnerable code, or produces a flawed ethical judgment — and you submit it — the error is yours alone.
Not the Passenger.
Senior Auditor
Every AI-generated output you use must pass through your expert judgment before submission. You verify logic, challenge claims, and correct errors — exactly as a compliance auditor would before signing off a report.
Decision Authority
AI may generate options. AI may synthesize frameworks. AI may write first drafts. The final strategic, ethical, or architectural decision must be yours — documented, reasoned, and defensible under oral questioning.
Disclosure Officer
Every major submission requires a documented AI Disclosure Appendix. You are the officer of record for the AI tools you used — listing tools, prompts, and the specific corrections or overrides you applied.
Ethical Guardian
In case studies and governance capstones, you bear sole responsibility for the ethical positions taken. AI cannot be cited as the source of an ethical judgment. That seat belongs to you.
This matrix governs all AI tool usage in graded assignments, capstone projects, and degree-program submissions. Authorized use accelerates your development. Prohibited use invalidates your credential — not because we are conservative about AI, but because a practitioner who cannot distinguish HITL-governed AI use from AI laundering is not qualified to govern AI systems in production environments.
Prohibited Use — Red Protocol
"Ghost" Submissions
Copy-pasting AI output directly into an assignment without substantial human editing, reasoning verification, and personal intellectual contribution. This is AI laundering — the most common and most severely treated violation.
Undisclosed AI Usage
Failing to document AI tools used in a submission via the mandatory AI Disclosure Appendix. Even authorized AI use, if undisclosed, constitutes a procedural integrity violation subject to enforcement.
Fabricated Citations
Submitting AI-hallucinated references, case law citations, or academic papers that do not exist. Every citation in every submission must be verifiable in the original source. Hallucinated references are treated identically to fabricated evidence.
Auto-Pilot Strategic Judgment
Allowing AI to supply the final ethical, strategic, or governance decision in a case study or capstone without documented human reasoning and override justification. AI outputs an option. You own the decision.
The AI Disclosure Appendix
Transparency is the first law of governance. Every major submission at RSIT must include a structured AI Disclosure Appendix at the end of the document. This is not bureaucratic overhead — it is the same audit trail structure that your future employer's AI governance officer will require before signing off on any AI-assisted decision in a regulated environment.
The Appendix must specify: which AI tools were used, for what purpose, and what human verification, correction, or override was applied to the output. This teaches you to produce the audit documentation that EU AI Act Article 13 and ISO 42001 require at the organizational level.
Submissions without a required Appendix are returned ungraded regardless of content quality.
═══════════════════════════════════════════ RSIT AI USAGE DECLARATION — REQUIRED ═══════════════════════════════════════════ SUBMISSION Module 4 Capstone — Data Governance STUDENT [Student Name] · [Cohort ID] DATE [ISO 8601 Date] ─────────────────────────────────────────── TOOL USAGE LOG ─────────────────────────────────────────── TOOL Claude 3.5 Sonnet (Anthropic) PURPOSE Python code generation for Data Quality validation module SCOPE Lines 12–47 of submission script ─────────────────────────────────────────── VERIFICATION RECORD ─────────────────────────────────────────── METHOD Manual test in AWS Sandbox ERRORS Logic error at line 40 identified CORRECTION NULL handling rewritten by student OVERRIDE AI schema recommendation rejected; replaced with GDPR-compliant design ─────────────────────────────────────────── COMPLIANCE STATUS ─────────────────────────────────────────── HITL STATUS ✓ COMPLIANT DISCLOSURE ✓ COMPLETE CITATIONS ✓ SOURCE-VERIFIED
Violations are reviewed by the Ethics & AI Committee using multi-modal detection heuristics, semantic pattern analysis, and oral defense verification. The sanction tier is determined by the severity of the violation and whether it constitutes a first occurrence or a pattern of behavior. There are no informal warnings for integrity violations. All determinations are documented and communicated in writing to both the student and their supervising faculty member.
Trigger: First-occurrence AI usage violation, missing Disclosure Appendix on a major submission, or minor undisclosed AI assistance that does not constitute systematic laundering.
Immediate failure of the specific module or assignment. Mandatory retraining on the RSIT AI Ethics Framework before the remediated submission is accepted. Module progress is paused pending completion of the remediation sequence.
Remediation RequiredTrigger: Pattern of AI laundering across multiple submissions, submission of fabricated citations, or deliberate circumvention of the Disclosure Appendix requirement with clear intent to misrepresent authorship.
Immediate revocation of RSIT LMS access. Full suspension of degree program progress. Formal notification issued to the student's supervising Academic Support Faculty. Reinstatement requires Ethics Committee review and written remediation plan approval.
Formal Review RequiredTrigger: Egregious, systematic, or repeat AI laundering after prior sanction; submission of fully AI-generated capstone work misrepresented as original research; or any integrity violation that demonstrates categorical unreadiness to govern AI systems in regulated environments.
Permanent expulsion from the Sherman Pathways ecosystem without refund of tuition or fees. Formal notification transmitted to all five partner universities. This determination is irreversible and will be documented in any credential verification requested by third parties.
No ReinstatementHuman-in-the-Loop — Every AI output is audited by you before submission. No exceptions. No auto-pilot submissions.
Full Transparency — All AI tool usage is documented in the mandatory Disclosure Appendix on every major submission.
Original Authorship — The reasoning, decisions, and ethical positions in your work are yours alone. AI assists. You govern.
Integrity is the Credential.
Questions about the Academic Standards Policy or the AI Disclosure Appendix requirement should be directed to your assigned faculty supervisor or to the Office of the Founder.