Reading the room.
Mapping the moves.

Cyber and digital policy spans hundreds of forums, files no one delegation reads together, shifts that arrive as linguistic drift in paragraphs nobody is monitoring, and venue-shaping patterns individually unremarkable and collectively decisive. A two-hour live panel testing what a small delegation can see, prepare for, and engage with when AI tooling is kept on a short leash and every claim cites its source.

Presented · 04 May 2026ParticipatoryCitation-backed~120 min

Three demonstrations · Two hours

Part 1·45 min·Mika Kerttunen (Aalto University)

Cyber-diplomacy simulation, live

Nine AI agents in prescribed national roles — US, EU, Russia, China, India, Brazil, South Africa, Singapore, UAE — working coalition formation around the Dutch proposal to protect the public core of the Internet. Each agent carries structured positions grounded in public corpora, not stereotypes. The capability the audience watches is linguistic drift: a delegation moving from 'we support' to 'we take note of', a co-sponsor quietly dropping off a paragraph — and the cascade implications surfaced live.

Audience moment

Audience works the same scene with the perception grid still on screen, anchored by: 'If this norm proposal landed on your desk tomorrow, what would your delegation's position be?'

Part 2·20 min·Chris Spirito (Sanctum, remote)

Strategic-engagement advisory

An audience-driven AI strategic-reasoning demo. Three normative ambitions — Norms in Action (Dutch-led), collective attribution practice, REAIM commitments on responsible military AI — × five regime-complex venues (UN First Committee, OECD AI, Freedom Online Coalition, regional bodies, multistakeholder forums). The audience picks one combination from their phones; the agent streams a coalition-building strategy with primary-source citation-back to a 58-document corpus.

Audience moment

Pick from your phone via the QR code below; tally builds live; the most-picked combination is what the agent reasons through on stage.

Part 3·40 min·Eneken Tikk

UN document positioning — where do the leads lie?

A taxonomy applied to documents.un.org to surface country positioning that's coherent across UN bodies whose files are not normally read together. All analysis pre-computed; every claim citation-linked to specific UN documents. Demonstrations for Russia, China, and a non-thesis country to show the instrument is country-agnostic. Also visible: the architects-coalition's frame survives into the Western normative track — the 2021 GGE consensus carrying 'sphere of information security' language verbatim.

Audience moment

Audience picks one country from a curated 9-country grid; that country's signature plays out, followed by: 'What pattern would you look for next if this were your file?'

The thesis under empirical test

Russia and China have become the architects of the global digital and cyber governance model — the baseline we need to push back against or deal with.

— Eneken Tikk's framing

This is not asserted rhetorically. It is shown on screen, with citations, and the audience inspects the evidence. The honest caveat is stated out loud, at the top: positioning-in-language is not the same as coordinated action. The tool tells us what countries say across files — which is consequential — but it is not the same as knowing what they do.

Participatory by design

Three explicit audience-engagement moments, one per part. These aren't polite flips — they're the substantive moments of the session. Everything else is in service of putting the audience in a position to engage from evidence rather than from priors.

Part 1 discussion — audience works the live scene with the perception grid on screen. Mika moderates; Eneken steps in for legal-normative questions.
Part 2 strategy pick— audience picks ambition × venue from their phones. The agent reads out the most-picked combination's coalition-building strategy with clickable citations to primary text.
Part 3 country pick— audience picks one country from the 9-country signature grid. That country's cross-body positioning plays out, followed by a reflection question.

Honest about AI, on purpose

Part of what makes this session worth attending is that it does not pretend AI assistance is magic. In a recent pilot, two of five test rows produced plausible-looking URLs that were confabulated outright. Stating this in front of a UNIDIR audience is deliberate. For a room of diplomats, methodology is the pitch; admitting the failure modes is the argument for trust. The rule we derived — every agent-returned URL is live-verified before an entry is promoted — is exactly the kind of operational discipline that separates useful AI tooling from decorative AI tooling. The same posture applies in Part 3: pre-computed analysis, strict citation-back, honest statement of what the tool sees and does not see.

Who is running the session

Finnish National Defence University · Aalto University

Mika Kerttunen

D.Soc.Sc. (Pol.), LTC (ret. Finnish Army) · cyber conflict prevention and rule-of-law-based resilience. Adjunct Professor, Military Strategy at FNDU; Visiting Researcher at Aalto School of Science. Consultant to UN GGE 2014–2015 (Estonia) and 2016–2017 (Finland); stakeholder representative, UN OEWG 2021–2025. Part 1 lead.

University of Tartu · TalTech, Estonia

Eneken Tikk

Dr.iur. · data, technology, and peace policy. Teaches data law and regulatory strategy at Tartu and TalTech; associate researcher at Helsinki University (international law) and Aalto University (AI and data science). Part 3 lead.

Idaho National Laboratory · Sanctum Security

Chris Spirito

Principal Investigator (INL) · Founder, Sanctum Security · AI safety, autonomous systems security, adversarial machine learning. Current NRC-funded research on cybersecurity for autonomous nuclear reactors, FPGA hardware vulnerabilities in safety-critical systems, and ML subversion in autonomous control. Co-author of 'The New Attack Surface' (with Rafal Rohozinski, SecDev Group); Co-Leader of NATO's Advanced Training Course Working Group 2. Part 2 lead (remote).

One-line takeaway

This is a session about what a small delegation can now see that it could not see before — and about doing this honestly enough that a UNIDIR audience can trust what they saw.