TITLE: Personal Memory Portability in the Digital Workplace (GDPR Amendment Proposal)
SUBTITLE: Individual Access to Digitally Mediated Work Records in an AI-Intensive Economy: Policy Options at EU Level
DATE: 2025-12-24T06:59:06+00:00
URL: https://npcmemo.substack.com/p/personal-memory-portability-in-the

======================================================================

Abstract

Work in the European Union is increasingly organised through digital systems that function as institutional memories, preserving not only outputs but also the contextual information required to interpret them, such as contribution histories, decision rationales, and review processes. Access to these systems is typically governed at organisational level and is commonly withdrawn at employment or project boundaries, with the result that individuals may lose practical access to the digitally mediated traces of their own work that are necessary for personal recollection and professional continuity, especially in AI-mediated environments.

This issue has become salient with the rapid diffusion of AI-enabled workplace systems. As organisations expand their capacity to retain and operationalise accumulated work records for automation, compliance, and institutional learning, asymmetries between organisational and individual access to work memory are intensified. Existing EU legal frameworks, including data protection and labour law, provide partial but insufficient coverage, as current rights of access and portability do not ensure a context-preserving record of an individual’s work contributions and must be balanced against intellectual property, confidentiality, and third-party rights.

The report examines the feasibility of recognising a narrowly defined “Right to Personal Memory Portability,” understood as a right enabling individuals to obtain a context-preserving personal memory record of their digitally mediated work contributions, subject to strict safeguards and exclusions. It assesses policy options at EU level, evaluates their legal and operational implications, and identifies areas requiring further empirical and legal analysis as digital and AI-mediated work practices continue to expand.

1) Problem Definition

Digitally mediated work is now produced, coordinated, and evaluated through enterprise platforms that function as systems of record: messaging and collaboration suites, ticketing and project-management tools, document repositories, design systems, code hosts, customer and service platforms, and security/audit layers. These systems preserve work outputs and also the surrounding context that makes outputs intelligible: decision rationales, approvals, review histories, issue discussions, and incident timelines. At the point of separation from an organisation (or a project boundary within it), access is commonly revoked at the identity/tenant level, such that the individual loses practical ability to retrieve their own work traces, even where a subset of those traces would be necessary for personal recollection, professional continuity, and the demonstration of acquired competencies.

This access loss is a loss of relational context. In many environments, the artefacts that support remembering (threads, comments, reviews, approvals, structured work histories, and cross-linked references) are not rendered portable in a meaningful way, and are often unavailable to individuals except while they remain within a given tenant. Eurostat indicators on remote access to enterprise resources illustrate the extent to which work is mediated by such systems, with a majority of EU enterprises offering remote access to core resources (email, documents, and applications), confirming that the “memory substrate” of work is increasingly platform-based rather than paper-based. European Commission

As AI and LLM-enabled software proliferate, the stakes of this asymmetry increase. First, these systems intensify the dependence on recorded traces as inputs to work (retrieval, summarisation, drafting, task routing), meaning that the ability to reconstruct prior decisions and learning is increasingly coupled to access to platform histories. Second, organisations can more readily convert accumulated work traces into institutional assets (knowledge bases, playbooks, automated support, analytics, and model-assisted processes), while individuals may remain unable to extract a commensurate “personal memory record.” Third, as AI governance regimes develop, auditability and traceability requirements will tend to increase organisational retention and logging, strengthening institutional memory while leaving the individual’s access position largely unchanged, absent an explicit policy intervention. The OECD’s recent work on AI in the workplace underlines the scale and immediacy of workplace adoption dynamics and the need for governance arrangements that ensure benefits depend on addressing associated risks. OECD+1

The policy problem, as framed for EU-level consideration, is therefore not a general claim of ownership over work product. It is the narrower claim that the conditions of digital work increasingly constrain personal agency through dispossession of one’s own work traces, and that this constraint becomes more salient as AI systems amplify the value of accumulated context and shape the informational basis on which future opportunities, performance narratives, and professional learning are constructed.

2) Legal and Policy Baseline

The current baseline offers partial coverage but does not clearly resolve the specific problem of context-preserving personal memory extraction from employment-related systems.

Under the GDPR, the right of access (Article 15) provides a mechanism to obtain confirmation and access to personal data, and data portability (Article 20) creates a right, under specific conditions, to receive certain personal data “provided by” the data subject in a structured, commonly used, machine-readable format, and to transmit it to another controller. EUR-Lex+1 The pre-GDPR lineage, through the Article 29 Working Party’s Guidelines on data portability (WP242 rev.01), emphasises that the purpose of portability is to empower data subjects and give them more control over their personal data, while also clarifying boundaries around scope and controller obligations. European Commission+1

However, the portability architecture is not designed to guarantee a portable record of work memory as such. Limitations include: (i) the requirement that portability applies where processing is based on consent or contract and carried out by automated means; (ii) interpretive debates over what counts as data “provided by” the data subject, particularly in collaborative systems where records are co-produced; (iii) the fact that much of what an individual needs for remembering is context and provenance rather than raw personal data fields; and (iv) practical constraints whereby exports omit thread histories, review context, and cross-linkage structures that are essential to reconstruction. The right to erasure (Article 17) is oriented to removal under defined grounds and exceptions and does not supply an affirmative continuity mechanism. EUR-Lex+1

Collision points arise with IP, trade secrets, confidentiality, and third-party data. The EU Trade Secrets Directive (2016/943) harmonises protection for undisclosed know-how and business information against unlawful acquisition, use, and disclosure, and is relevant insofar as many work traces embed confidential business information, client details, and organisational methods. EUR-Lex+1 Even where a record contains personal data, it may simultaneously contain protected trade secrets or third-party personal data, requiring a balancing and redaction logic rather than a binary entitlement.

A further baseline consideration is the labour-law environment. EU-level instruments establish minimum requirements regarding information and consultation (Directive 2002/14/EC) and other working-conditions transparency obligations (Directive (EU) 2019/1152), but these do not amount to a harmonised right of post-employment access to digitally mediated work traces. EUR-Lex+2EUR-Lex+2 National employment record practices vary substantially, including the role of worker representation structures (e.g., works councils) and sectoral regimes, limiting the effectiveness of purely labour-law approaches for cross-border platform environments.

3) Definition and Scope of Right to Personal Memory Portability (RPMP)

The proposed Right to Personal Memory Portability (RPMP) is conceived as a narrowly scoped, EU-level right enabling an individual to obtain a context-preserving “personal memory record” of their digitally mediated work contributions for purposes of personal recollection and professional continuity, without creating a general right to reuse, disclose, or commercially exploit protected information.

A “personal memory record,” for RPMP purposes, would be defined as a structured bundle comprised of the following components, insofar as they are technically available and can be provided subject to safeguards.

First, authored content: records created by the individual in the course of work, including text, code, designs, and other authored artefacts, to the extent that provision does not disclose excluded material.

Second, contribution metadata and provenance: timestamps, authorship markers, contribution summaries, and other provenance indicators necessary to establish that the individual materially participated in the creation or modification of artefacts, including contribution histories in versioned systems, and work item participation markers.

Third, decision context: the subset of surrounding context necessary to render the authored content intelligible, including comment threads on the individual’s contributions, review feedback addressed to the individual, approval states relevant to the individual’s work items, and cross-references (e.g., linking a work item to a change request or incident) to the extent required for coherence.

Fourth, outputs generated or materially shaped by the individual: where the individual’s role is primarily curatorial or coordinative, the record may include structured “participation traces” that evidence substantive shaping (e.g., authored decisions, approvals, meeting notes) rather than the entire underlying collaborative corpus.

RPMP would include explicit exclusions and constraints. It would not provide: trade secrets and sensitive business information beyond what can be safely abstracted; sensitive client data; third-party personal data not necessary to the individual’s recollection; privileged legal material; regulated security data (including certain logs and threat intelligence); or materials subject to legal holds where disclosure is prohibited. The right is oriented to remembering and continuity and must be framed with purpose limitation and controlled disclosure as central features, rather than assuming that extraction entails unrestricted possession.

4) Options Appraisal

Option A: Interpretive guidance within existing GDPR rights. Under this option, EU-level regulatory guidance (through the EDPB and national DPAs, consistent with the WP29 lineage) would clarify that, in employment-related contexts, access and portability should be interpreted to include a minimally sufficient “context layer” for intelligibility, subject to redaction. The principal advantages are speed and lower legislative overhead; it would build on established rights and supervisory practices. The limitations are material: interpretive expansion would remain contested, unevenly enforced, and constrained by the GDPR’s scope constructs (notably “provided by,” lawful-basis conditions for portability, and third-party/confidentiality balancing). The risk of fragmentation and litigation is significant, and it may not provide the clarity necessary for standardised implementation across platforms.

Option B: Legislative establishment of a distinct RPMP right within EU data protection architecture. This option would introduce an explicit, narrowly framed right that is adjacent to but distinct from Article 20 portability, specifying the notion of a “personal memory record,” defining scope and exclusions, and requiring implementable safeguards. The advantages include harmonisation, definitional clarity, and the ability to explicitly encode proportionality tests and redaction obligations. The burdens include legislative complexity, compliance costs for controllers, and the need for careful interaction with trade secrets protection and third-party rights. Feasibility would depend on demonstrating subsidiarity and proportionality and articulating a credible enforcement and standardisation pathway.

Option C: Labour-law–adjacent instrument or sectoral approach. This option would pursue an employment-focused right, potentially via amendments or new minimum requirements addressing post-employment access to certain categories of work records, with a strong role for social partners and worker representation. This could align with existing labour-law rationales and address workplace power asymmetries directly. However, fragmentation risk is higher given national variance, and the approach may not cover non-standard work arrangements, cross-border platform use, and contractor/project-based labour that is increasingly characteristic of digital work. A sectoral approach could be feasible for high-digitisation sectors but risks leaving general knowledge workers without coverage.

Across all options, enforceability and abuse risk hinge on the same engineering problem: specifying what is extractable, under what safeguards, with what redaction, and through what auditable process, such that the right is usable for remembering without functioning as a general exfiltration channel.

5) Implementation and Safeguards

Operational implementation would require practical mechanisms that treat RPMP as a controlled, auditable extraction right rather than a bulk export.

Controllers would provide secure portals enabling individuals to request a personal memory record in standardised formats. Exports should preserve core structural relationships (e.g., linkages among work items, authored artefacts, and review states) sufficient for intelligibility. Redaction workflows would be required to remove excluded material, including third-party personal data and protected trade secrets, with a documented rationale.

Time-bounded post-employment access is a plausible implementation pathway where immediate export is impractical. For example, a secure, read-only access period limited to extraction of the personal memory record could reduce implementation complexity while limiting risk. Audit logs should record what was extracted and under which rules. Controllers should provide an appeals and complaint pathway, including escalation routes consistent with supervisory authority structures in data protection contexts.

Mitigation for exfiltration risk would require least-privilege extraction, purpose limitation, and technical and organisational measures. These may include tiered extraction (starting with authored content and provenance, adding context only where necessary), rate limiting, watermarking, and contractual attestations regarding non-disclosure. The Trade Secrets Directive baseline makes it necessary that acquisition/use/disclosure constraints remain intact; accordingly, RPMP must be designed to avoid creating legal ambiguity about lawful acquisition and permissible uses. EUR-Lex+1

Standardisation should be approached as an interoperability programme rather than vendor-led productisation. A schema-based “memory record” specification, developed through EU standard-setting channels and stakeholder consultation, would reduce administrative burden and support cross-platform portability without implying that all underlying system data is extractable.

6) Recommendations and EU Strategy

A first recommendation is to commission a feasibility study and impact assessment focused on definitional engineering, technical viability, administrative burden, and cross-border enforcement, explicitly distinguishing RPMP from existing portability rights and mapping interactions with trade secrets and third-party rights.

A second recommendation is to pilot RPMP-like extraction mechanisms within EU institutions and agencies as controlled testbeds, using real systems-of-record and procurement constraints to surface practical implementation issues and costs.

A third recommendation is to convene structured engagement with DPAs, social partners, employer associations, and platform vendors to develop a workable taxonomy of extractable “personal memory record” components and a shared redaction/proportionality logic, building on the WP29/EDPB tradition of guidance development. European Commission+1

A fourth recommendation is to initiate a standards trajectory for context-preserving export schemas and auditability requirements, aligned with existing security and compliance expectations, and mindful of the realities of remote, platform-mediated work. European Commission

A fifth recommendation is to ensure coordination with EU AI governance work, in particular where workplace AI systems increase retention, logging, and traceability, and where model-assisted processes may intensify asymmetries between institutional and individual access to work traces. Where precise legal obligations under AI governance instruments are relevant, they should be incorporated cautiously, with an emphasis on traceability and accountability dynamics rather than speculative claims about model training.

A sixth recommendation is to develop a guidance note on evidence and dispute contexts: how an individual may use a personal memory record for self-explanation and continuity while respecting confidentiality, and how controllers may rely on auditability to demonstrate that extraction did not constitute unlawful disclosure.

7) Limitations and Open Questions

The highest-risk ambiguities concern collaboration and mixed authorship. Many relevant artefacts are co-produced, and deciding what portion is necessary for one individual’s remembering without disclosing others’ personal data or confidential material remains non-trivial.

AI training traces and prompt histories present further uncertainty. Where work is mediated through AI tools, the boundary between the individual’s authored contribution and system-generated output may be unclear, and the degree to which prompts, outputs, and usage logs are included within a personal memory record requires careful design.

Jurisdictional variance in employment law affects feasibility for labour-adjacent approaches, particularly regarding post-employment access rights, record retention, and the role of worker representation.

Finally, enforcement outside standard employment relationships (contractors, gig workers, cross-border project labour) remains an open question. Any EU-level approach would need to specify the scope of covered relationships to avoid excluding precisely the forms of work most dependent on platform-mediated memory.

Overall, the RPMP concept appears feasible only insofar as it is defined narrowly, engineered for redaction and auditability, and positioned as a continuity right compatible with existing protections for trade secrets, security, and third-party rights, rather than as a general claim over organisational information.
