Private AI Use Doctrine

The position

PLENA's position rests on three architectural commitments. They are not loopholes and they are not permission to break rules. They describe where private process ends and legitimate disclosure begins. Most existing AI-ethics infrastructure assumes disclosure is virtuous and detection is legitimate enforcement. PLENA takes the contrary view: AI assistance is becoming ordinary, the human stays responsible for the final work, and exposure of private process should be the exception that a real rule justifies — not the default that everyone is presumed to owe.

Private process stays private

Private process includes prompts, notes, drafts, failed versions, outlines, and experiments — except where a valid rule, contract, court order, or professional duty requires specific disclosure. The messy interior of how a person reaches a finished work has never been something they owe the world by default. A writer's discarded paragraphs, a researcher's dead-end queries, a student's rough outline: these are working material, not public record, unless a legitimate rule actually says otherwise.

The human stays responsible

A person can use AI tools and still remain the responsible author, student, worker, researcher, lawyer, priest, or creator. Tool use is not the morally significant fact; human responsibility for the final work is. The question that matters is not "was a tool involved?" but "who stands behind the result?" PLENA's instruments let a person answer the second question clearly without being forced to surrender the first.

Disclosure stays proportionate

When disclosure is required by a legitimate authority with a legitimate basis, the user should disclose the minimum the rule actually requires — not the entire private process. A rule that asks "did you use AI within permitted limits?" is not a rule that asks for every prompt, every draft, and every search. Proportionate disclosure satisfies the legitimate need without converting a reasonable question into total exposure.

The framing

AI assistance is becoming infrastructure — like spellcheck, calculators, dictionaries, search engines, editors, tutors, and research assistants. We do not ask writers to disclose every dictionary lookup, students to log every spellcheck correction, or researchers to attach a transcript of every search query to their results. Demanding full disclosure of AI use is increasingly equivalent to those demands: technically possible, rarely proportionate, and corrosive to the ordinary privacy of thought.

The human remains responsible for the work. The tools used to produce it are private process unless a legitimate rule actually requires otherwise. This is not a claim that anything goes; it is a claim that the burden of justification belongs on the demand for disclosure, not on the ordinary use of a tool.

The taglines

Read these as design principles, not as marketing copy. The primary line anchors everything; the others are the same idea in the register of the audience hearing it.

The three problems PLENA addresses

1 · Data exposure through AI tools

Users — especially professionals such as lawyers, doctors, therapists, priests, teachers, researchers, and business operators — may share sensitive client, patient, student, parish, or business information with AI tools without realizing the disclosure implications. A prompt pasted in haste can carry confidential material into systems the user does not control. PlenaProof helps users check what they may have exposed and what disclosure obligations may follow — and record that they ran the check.

2 · False AI accusation

Users — especially students, writers, researchers, and journalists — may be wrongly accused of AI use based on AI-detection results. AI-detection results can be contested, especially when used alone. PlenaProof helps users assemble process evidence, policy context, and appeal records without surrendering private creative records. The goal is not to "beat a detector"; it is to let a wrongly-accused person show their process on their own terms.

3 · Disproportionate disclosure demands

Some institutions demand full prompt histories, draft transparency, or blanket "no AI" attestations beyond what a legitimate rule actually requires. PlenaProof helps users document the demand, the user's response, and the basis for any limited disclosure offered — so that an overreaching demand and a proportionate reply are both on the record.

The privacy-safe architecture principle

PLENA's defense of private AI use must itself respect the rights it defends. A tool that protected privacy by collecting private process would be a contradiction. So the architecture is built to prove facts about private process without ever requiring the private process to become public:

This is the architectural commitment that makes "Private AI use. Public human responsibility." substantive rather than rhetorical. PLENA documents facts about private process without ever requiring the private process to become public. The receipts that follow from this doctrine — the Human Responsibility Receipt, the AI Use Privacy Check Receipt, the Disclosure Demand Receipt, and the Private Draft Receipt — all sign proof structure, never content.

The four receipt types

Four additive PLENA VRX-1 receipt types carry this doctrine into practice. Each signs proof structure only — never drafts, prompts, or AI outputs. The schemas are published as additive extensions to the VRX standard:

• Human Responsibility Receipt — attests to ownership of the final work without exposing private process.
• AI Use Privacy Check Receipt — attests that a privacy check was run against named risk categories.
• Disclosure Demand Receipt — records a disclosure demand and the user's response.
• Private Draft Receipt — content-free, privacy-safe proof of authorship process.

These power the PLENA Private Augmentation Shield, the sector suite that packages the position for individuals and institutions.

For institutions

PlenaProof helps institutions avoid crude AI panic. Blanket disclosure demands and detector-only enforcement create false-positive harm, equity problems, and legal exposure. PLENA's institutional infrastructure replaces these with proportionate disclosure, documented appeal pathways, process receipts, and privacy-safe review. Institutions that adopt PLENA's posture demonstrate seriousness about integrity without imposing surveillance that punishes legitimate users.

In the institutional register: disclosure when required, privacy by default, responsibility always. PlenaProof complements an institution's integrity processes; it does not replace the institution's judgment, its rules, or its authority to set them.

The architecture that makes this possible

The Private AI Use position is structurally available to PLENA because PLENA depends on no single centre of AI infrastructure. PlenaProof is planetary trust grammar — a shared receipt format that any institution, partner, or court can verify, owned by no single platform. Privately augmented work is part of that grammar: a person anywhere can stand behind their work without routing their private process through, or depending on, any one platform's surveillance expectations.

The equity and accessibility connection

Forced AI-use disclosure does not fall evenly. It may disproportionately affect non-native English speakers, dyslexic writers, neurodivergent students, diaspora writers maintaining heritage languages, accessibility-need users, and second-language academics — precisely the people most likely to use drafting, translation, and editing assistance to participate on equal terms. A regime that treats tool use as suspicion punishes the people for whom the tools are most enabling. PLENA's defense of private AI use functions, in part, as accessibility infrastructure: it lets people use the help they need and still be judged on the work they stand behind.

The truth boundary

School and publisher policy boundary

PLENA speaks in receipt grammar, not verdict grammar: it records what was attested, demanded, checked, and responded to. It does not rule on truth, and it makes no promise of any specific outcome with any specific institution. AI creates; PlenaProof proves the record.