Cyber sovereignty is on ITBN's agenda. We did not put it there.

The event

ITBN CONF-EXPO 2026 runs on 24 September 2026 at the Bálna Honvédelmi Központ in Budapest. It is the twenty-second edition of an event that has been the Hungarian information security profession's September fixture since 2005. The organisers put the floor at 5,300 m², up from 4,200 m² last year, and expect more than 1,500 participants.

The motto this year is "Young and Vets, hAil the threats!" — and the capitalisation is deliberate. The AI sits inside hAil, and it points at autonomous attacks generated by machines rather than typed by people.

What we find more interesting than the motto is the plain-language description of who the event is for. In the organisers' own Hungarian: everyone who wants to act "a kibervédelemért, az AI kockázatainak csökkentéséért, a kiberszuverenitás védelméért és az ellátási láncaink biztonságáért" — for cyber defence, for reducing the risks of AI, for the protection of cyber sovereignty, and for the security of our supply chains.

We have been making the sovereignty argument to Hungarian companies for two years and it has often landed as a vendor's talking point. It is a different thing when the profession writes it into its own conference description.

NIS2 stopped being a principle and became a number

The other change worth noticing is smaller and much more concrete. NIS2 has been discussed in Hungary as a direction of travel for years. This year's ITBN carries a national meeting for information security officers built around regulation 17/2025 (VII. 24.) EM and its twenty hours of mandatory training per year, with an expo floor of accredited training providers attached to it.

Twenty hours a year per officer is a budget line. It has a cost, an owner and a deadline. That is a different kind of conversation from "cyber resilience": it is the kind that survives contact with a CFO.

Shadow AI is the security problem of this cycle

The agenda item we expect to be most argued over is Shadow AI — staff putting company material into public AI tools that nobody approved, nobody logs, and nobody can retrieve from afterwards.

It is a genuinely hard problem because it is not a technical failure. It is what happens when a capable tool is one browser tab away and the sanctioned alternative is worse or does not exist. The employee pasting a draft contract into a public chatbot is not being reckless. They are being efficient with the tools they were given.

This is the point where our own interest is obvious, so let us state the limit before the argument.

Why the timing is not a coincidence

The same week ITBN was finalising an agenda about AI-generated attacks, Broadcom's VMware division published its own account of what is reshaping enterprise infrastructure, and named the same force from the other direction: frontier models are accelerating the discovery of vulnerabilities, so that, in their words, "attacks can happen in hours not weeks".

That is the part of the AI story that reaches a security team first. Not whether a model becomes superintelligent, but that the tooling on both sides of the fence got faster at the same time, and the defender's side of it is bounded by how quickly a human can approve a change.

We will be at ITBN. If you are going and want to argue about any of the above, that is the better use of the day than a demo.