Expert-supervised attribution
Use inventors and domain experts to evaluate technical relationships, with reasons, uncertainty, precise evidence spans, and independent second review. Technical judgments remain separate from legal determinations.
Trace the contribution. Govern the use.
Can expert-grounded attribution and traceable rights evidence protect IP while keeping legitimate work usable?
AI-assisted work draws on code, documents, models, retrieval, and human edits. A technical similarity finding cannot, by itself, establish how a contribution arose or whether its use is authorized. My research connects expert-supervised attribution with a separate governance process for access, transformation, and release.
Use inventors and domain experts to evaluate technical relationships, with reasons, uncertainty, precise evidence spans, and independent second review. Technical judgments remain separate from legal determinations.
Study how provenance can remain useful across human edits, model responses, retrieval, tool use, branches, merges, and handoffs. Compare recorded lineage with independently observed creation events.
Connect attribution findings to separately resolved ownership, license, permitted use, and release decisions. The question is whether the combined process protects IP without unnecessarily blocking authorized work.
The existing Spyris platform and extension provide a starting point for research. Controlled experiments and consented workflows can test evidence capture, expert agreement, appropriate reliance, and the quality of access and release decisions.
This program includes company research papers, technical proposals, and experiment designs. Proposed invention families and contractual background are tracked separately. No experimental accuracy, legal determination, or current enforcement guarantee is inferred from a draft.
Research framing for attribution and the transition of technology between organizational settings.
Connects evidence about technical relationships with review, reuse, and transition decisions. Companion document formats represent one work.
A framework for considering innovation, licensing, and deployment across organizational boundaries.
Separates background and foreground IP and identifies the role of authorized review in deployment. The paper does not itself grant rights or establish compliance.
An early feasibility record containing a small synthetic demonstration and review conditions.
Provides development history for the evaluation program. A twelve-query synthetic demonstration is not treated as a general performance benchmark or field validation.
An architecture and proposed evaluation plan for attribution and rights-aware decisions.
Connects hybrid attribution methods with policy requirements and evaluation planning. A proposal is not evidence of a contract award or measured deployment outcome.
A distinct revision preserved in the research history.
Records the evolution of the proposal without assuming every claim in a later draft supersedes the evidence in earlier sources.
Research framing for attribution evidence and the organization of intellectual property.
Connects attribution with pooling and authorized reuse. The public summary covers the research framing; private correspondence included in the source is excluded.
A partner-oriented framework connecting attribution, expert validation, governed delivery, custody, and telemetry.
Develops the relationship between attribution findings and the operational decisions surrounding IP. Demonstration data is not presented as production performance.
A proposed controlled process for collecting expert judgments and workflow evidence.
Frames how expert assessments can support training and evaluation while retaining uncertainty, independent review, and separation between source families.
The integrated research agenda prepared for academic discussion.
Proposes comparing attribution alone with attribution plus governance, using unauthorized disclosure, mistaken approval, unnecessary blocking, custody reconstruction, and legitimate utility as outcomes.
Attribution alone versus attribution with governance; separate training and evaluation by source family and inspect model outputs directly.
Expert agreement, evidence quality, calibrated uncertainty, unauthorized disclosures, incorrect approvals, unnecessary blocks, and legitimate task completion.
Evaluate whether a reviewer can reconstruct custody, identify the exact asset version and recipient, and confirm that revocation or release decisions took effect.
Evidence of a technical relationship, authority to act, and proof of an outcome are different things. The research keeps each visible.
This public research profile was prepared from my research portfolio and source register dated September 21, 2026, together with the linked publication records. Dates shown as source-record dates identify the available version; undated work remains undated.
Working-paper summaries are linked above. Full collaboration materials are available for appropriate research review. Signed agreements, private correspondence, and confidential invention disclosures are not public downloads. Licensed NSA background technology remains distinct from my and my companies’ research contributions.
Research collaborations across national security, civilian institutions, and human health.
Discuss the research