Engineering Cross-Compatible Personalization for Interactive AI-Driven ServicesEngineering Cross-Compatible Personalization for Interactive AI-Driven ServicesEngineering Cross-Compatible Personalization for Interactive AI-Driven ServicesEngineering Cross-Compatible Personalization for Interactive AI-Driven Services

Workshop @ EICS 2026, 30 June - 3 July 2026, Patras, Greece

Workshop Website: https://ecpai26.netlify.app/

Introduction and Motivation

Contemporary interactive systems increasingly embed AI components that personalize services, interfaces, and decision workflows. These systems build and maintain computational user representations in terms of inferred preferences, intents, and behavioral data, which, while impacting opportunities and experiences offered to users, yet remain largely opaque, fragmented across platforms and hard to update by them. Users are thus compelled to sacrifice privacy, or accept platform lock-in as preconditions for receiving personalized services.

This situation raises a pressing engineering challenge for the EICS community: how do we systematically design architectures, tools, and interaction techniques so that personalization becomes negotiable, inspectable, and portable? Previous research on scrutable adaptation and user-controllable personalization has highlighted the engineering difficulty of making AI-driven user representations transparent and editable. Indeed, transparency mechanisms in deployed systems remain predominantly reactive: they often offer post-hoc explanations of system behaviour, rather than proactive support through which users can inspect, contest, and revise their digital representations before personalisation decisions are made.

This challenge is further emphasised by emerging regulatory obligations under the GDPR and the EU AI Act, which impose obligations of transparency and meaningful user oversight on AI-driven services. Meeting these obligations demands not merely legal compliance, but the careful engineering of interactive systems whose AI components should be not only context-adaptive, but also controllable and auditable.

Objectives

This full-day workshop brings together researchers and practitioners from interactive systems engineering, human–computer interaction, AI/machine learning, and security/privacy to address the engineering of cross-compatible personalisation, a vision for interactive AI-driven services in which users can make their digital self-representations inspectable, editable, and selectively shareable across different platform boundaries.

The workshop is structured around three main objectives:

(1) Establish a shared vocabulary and design space for negotiable personalisation, integrating perspectives from interaction design, AI engineering, and privacy and security research.

(2) Produce reusable engineering artefacts: a reference architecture for user-sovereign personalisation systems, an interaction pattern catalogue for user model inspection and editing, a minimum viable cross-platform profile negotiation protocol.

(3) Build a cross-community research network spanning interactive systems engineering, user modelling , usable privacy and security, and decentralised identity infrastructures

Topics 

We solicit contributions on the following (non-exhaustive) list of topics:

• representations for user-controlled personalization (e.g. schemas, modular profiles, local adapters);

• interaction techniques for inspecting/editing user models (e.g. dashboards, timelines, profile composition tools, conversational agents, what-if simulation and preview mechanisms);

• negotiation workflows and consent mechanisms (e.g. time-boxing, selective disclosure, revocation, audit trails);

• architectures for user-side AI (e.g. on-device/federated learning, privacy-preserving exchange);

• verifiability and accountability (e.g. credentials, provenance, logging, evaluation of trust and UX);

• empirical methods and benchmarks for measuring agency, predictability, and utility.

Submission Guidelines

We invite the following submission types:

• Position papers (2–4 page PDF in ACM Manuscript format, references excluded): describing research results, open engineering problems, or system visions relevant to the workshop scope;

• Artifact sketches (2-4 page): presenting an interface concept, protocol or API sketch, or architectural diagram;

• Optional supplementary material: a short demonstration video or repository link may accompany any submission to support workshop discussion.

Submissions must be uploaded via EasyChair at https://easychair.org/conferences/?conf=ecpai2026. All submissions will be reviewed by the programme committee for relevance, scientific quality, and potential to stimulate discussion. Accepted contributions will be presented as lightning talks and discussed in structured breakout sessions. Upon acceptance, at least one author is required to register for the workshop, and attend the workshop in person. 

Post-workshop plan

We will invite the authors of accepted submissions to revise and extend their work following the workshop discussion for inclusion in a joint LNCS post-workshop volume published by Springer.

Important Dates

Submission deadline 25 April 2026, 23:59 AoE

Notification to authors 2 May 2026

Camera-ready for workshop website 20 May 2026

Workshop day 30 June 2026