Media Brief · Governing Relationship State
What journalists need to know
Micheal Charles Preble · Independent Researcher
Why this research matters
An AI system can hold information about you that was true when recorded but is no longer an adequate basis for what it's currently being used to decide. This research proposes a legal framework for exactly that gap — not deletion, not permanent retention, but governed continuing reliance.
The research in one sentence
The relevant legal question for persistent AI memory is not who owns it, but whether continuing reliance on it is still lawfully justified — and the paper proposes eight duties (visibility, contestability, provenance, revalidation, restriction, retirement, transition review, and safeguards against improper reconstitution) that should attach to whoever controls that reliance.
What is new
A framework that treats “Relationship State” as potentially governed by different legal interests simultaneously, and introduces “Material Reconstitution” — the idea that recreating substantially the same operative state from summaries or derivative records after it was corrected or retired should trigger the same review as if the original state had returned.
What is established vs. proposed
Established (real, cited law): GDPR Articles 18–20 already separate storage from use and portability from ownership; the FCRA already requires reinvestigation and conditions reinsertion of deleted information on certification and notice.
Proposed by this paper: the synthesis of these existing mechanisms into a purpose-limited governance interest specific to persistent AI relationship state. This is legal-institutional analysis, not new empirical data.
What would be inaccurate to say
- That this paper proposes a new ownership right in AI memory. It explicitly rejects ownership as the master variable.
- That the framework requires deleting information whenever its authority is uncertain. It explicitly prefers restriction or retirement over deletion in most cases.
- That the paper has identified how often this problem actually occurs in deployed systems. It does not measure prevalence.
- That a single institutional vehicle (a new federal law, for instance) has been shown to be the right implementation. That question is explicitly left open.
Canonical source / citation
Read the canonical paper record → · Full paper PDF · View on SSRN
Preble, Micheal Charles. “Governing Relationship State: A Legal Framework for Human-Controlled AI Continuity” Available at SSRN 7396659, 2026.
Researcher contact
Micheal Charles Preble · Independent Researcher · micheal@perfinitive.com · ORCID
No quotations are pre-approved for press use on this page. Contact directly for comment or an approved quote.