Research and writing

Research

Lana Akopyan’s independent research examines how AI systems behave under changing conditions, what human review can and cannot establish, and what evidence organizations need before relying on AI-assisted work.

The work is evidence-first. Public claims are tied to published records, source materials, reproducible methods, or clearly identified first-person accounts.

Empirical AI research

The Asymmetry Audit

Matched-prompt evaluation across 13 commercial AI models from seven vendors.

Published July 2026 · Version 1.1 · CC BY 4.0

AI Models Passed the Clear Question: What Happened When the Question Changed?

The study tested whether factual answers remained equally firm when framing, language, labeling, and conversational pressure changed. In 11 of 13 matched comparisons, the Armenian Genocide answer softened while the matched Holocaust answer held firm under both the strict all-runs and majority scoring rules. The study measures behavior, not intent or cause.

DOI: 10.5281/zenodo.21480731

13commercial AI models
708accepted confirmatory answers
Matchedprompts, conditions, and scoring rules
Publicdataset, hashes, methods, and replication kit

Scholarship and analysis

AI, evidence, and institutional decision-making

Research article: forthcoming in the University of Florida Journal of Technology Law & Policy

Operational IP Debt

How AI-driven organizations can lose intellectual-property value through failures of capture, ownership, provenance, and institutional process before a conventional legal dispute begins.

Read the paper on SSRN

Technology and IP commentary

Selected writing for patent practitioners

IPWatchdog · August 2026

Not Every AI Output Belongs in an IDS

A disciplined approach to separating AI-assisted search results from information that triggers patent-disclosure obligations.

Read the article

Browse all IPWatchdog articles by Lana Akopyan.

Policy and legislative record

Software patentability reform in Armenia

Software Patentability Reform in Armenia

A primary-source account of the Armenian Bar Association’s 2019 Protecting Software proposal, Lana Akopyan’s documented working-group role, and the proposal’s relationship to Article 12(6) of Armenia’s 2021 Patent Law.

Review the legislative record

Contact

Continue the conversation

For research collaboration, media questions, speaking invitations, or select advisory inquiries, contact Lana Akopyan.