Dated record
News
What changed, when, with the public anchor that lets a reader check it: a DOI, a registry entry, a programme page, a decision date. Items are written when the fact is verifiable, not when it is hoped for. For the standing picture see the track record; for field reports see Expeditions.
2026
-
Publication
The AI export-control article is out in Frontiers in Political Science
What precise enforcement of an AI export control needs at the point of use, read through the June 2026 suspension of two frontier systems and the 2017 Estonian ROCA recall; volume 8, article 1908543, DOI 10.3389/fpos.2026.1908543.
-
Public engagement
Invited talk at the Tallinn Kristiine Rotary Club: Before the Agent Acts
A 30-minute talk for a general audience: 4 questions to ask before an AI agent pays, books or signs for you, a wallet prototype that shows the person the exact action, and who can check the record afterwards.
-
Publication
The MIRROR article is out in the Journal of Open Research Software
Local evidence bundles that keep each claim of a research article next to the files that support it, with an offline verifier and the JANUS case study, published in volume 14, article 65, DOI 10.5334/jors.753.
-
Publication
The patent-corpus reproducibility article is out in World Patent Information
A repair-first pipeline for patent evidence bases and 13 reporting items for the USPTO Open Data Portal transition, published in volume 87, article 102511, DOI 10.1016/j.wpi.2026.102511.
-
Open data
An open dataset of the EU trusted lists, as 5 instruments saw them
14,539 fetch attempts and 896 document versions of the EU's trusted lists, 2010–2026, with a methods note: the 27 collapses in the daily provider total were failed downloads, not providers leaving.
-
Teaching
Guest lecture at TalTech: Before the Agent Acts
A 60-minute guest lecture with a live quiz in the course Digital Society Principles: Towards e-Governance: when a correctly signed action is still outside its mandate, and what a verifier reads before the agent acts.
-
Research networks
COST Action CA25178 holds its first Management Committee meeting
The Action on agentic physical AI started in Brussels: officers elected, five working groups set up, and an evidence-and-conformance task carried by Estonia's Management Committee member.
-
Publication
The SD-JWT differential-fuzzing paper is out in Springer LNCS
Spec-as-(N+1)-oracle differential fuzzing of SD-JWT presentation verification, presented at STM 2026 in Rome, is published in Lecture Notes in Computer Science.
-
Peer review
Sole Control for Machines accepted for the NGISE 2026 proceedings
An eIDAS 2.0 census of remote qualified signature-creation devices and a trust primitive for accountable agentic information systems, accepted for Springer's CCIS series.
-
Peer review
MIRROR accepted by the Journal of Open Research Software
Local evidence bundles for reproducible research-software review, with a JANUS case study, accepted after three rounds of review.
-
Publication
When the Witness Is Also the Suspect, in IEEE Internet Computing
The case for hardware-rooting the evidence that AI agents leave behind is online as an early-access article.
-
Funding
A research project co-funded by the European Union
The institute's research project on accountable AI-agent actions is co-funded by the European Union from cohesion policy funds.
-
Publication
XATF published in IEEE Access
A carrier-binding verification profile with typed failure attribution for AI-agent evidence over X-Road, open access under CC BY.
-
Publication
Agent-access receipts published in Frontiers in Computer Science
A reproducible synthetic security experiment on what an access decision for an AI-mediated caller should leave behind.
-
Membership
Tyche Institute joins the Agentic AI Foundation as an Associate Member
An Associate Member of the Linux Foundation's agentic-AI fund, the home of MCP, A2A, goose, AGENTS.md and agentgateway. No vote, no board seat.
-
Standards
Mapping the PQC Maturity Model onto the eIDAS/ETSI normative base
A crosswalk between the PKI Consortium's Post-Quantum Cryptography Maturity Model and the ETSI/CEN standards EU trust service providers are audited against.
-
Standards
A registered media type for action evidence
IANA registered application/vnd.aep+zip, the media type for an Action Evidence Package, with Tyche Institute as change controller.
-
Policy
Tyche Institute files a position with the European Commission on the Cloud and AI Development Act
European AI infrastructure should be required to produce verifiable, portable evidence of automated action. Calibrated against an open corpus of 2,244 consultation responses.
-
Research networks
Estonia appoints Anton Sokolov to the Management Committee of COST Action CA25178
A national-representative appointment to the Action on agentic physical AI, with Tyche Institute as the affiliation.
-
Governance
Building independent research-ethics review at Tyche Institute
A two-reviewer, arm's-length process for minimal-risk human-participant research, alongside review by Estonia's Research Ethics Committee where appropriate.
-
Publication
Nekropolis is now published in Data in Brief
The open-access data article describing a 14,923-record corpus of computer-science retractions and AI-governance lifecycle records.
-
Standards
Our first IETF Internet-Draft
Composing an AI agent's Action Evidence Package with hardware remote attestation under RATS (RFC 9334).
Notes
-
Speaker notes
Agent-action evidence and remote attestation
Notes and figure from the IETF 126 side meeting on remote-attestation challenges for AI agents.
Items name only what a public record supports. Appointments are stated as individual roles where they are individual; acceptances are stated as acceptances until the publisher's record exists. Nothing on this page is an endorsement of Tyche Institute by the bodies named.