June 5–6, 2026 · Nuremberg, Germany

AI Transparency Conference

AITC 2026 has concluded.

Thank you to everyone who joined us in Nuremberg — speakers, authors, reviewers, and attendees. This page remains online as an archive of the programme. Planning for AITC 2027 is underway — everything will be posted at aitc2027.cc.

Impressions from AITC 2026 — two days in Nuremberg.

Greetings from IASEAI, co-sponsor of AITC 2026.

cosponsored by

Keynote Speakers

Vera Schmitt

Research group lead
XplaiNLP Group, TU Berlin

Joel Z. Leibo

Senior staff research scientist
Google DeepMind

Yonatan Belinkov

Harvard & Technion

The AI Transparency Conference is an international research conference focused on advancing transparent and human-compatible AI systems. It brings together researchers working on interpretability, AI safety, control, and governance.

The 2026 edition was held on-site in Nuremberg over two days, bringing together PhD students, postdoctoral researchers, faculty, and research institutes for three keynotes, ten paper talks, and two poster sessions. Attendance was free of charge.

When & Where It Took Place

Friday, June 5, 2026 12:00 – evening

Saturday, June 6, 2026 09:00 – evening

Südwestpark 97 · 90449 Nürnberg

See the full programme below.

Talk Recordings Full Programme AITC 2027

The conference addressed three core themes

Detect

Agent Behaviour & Collaboration

Methods for observing, measuring, and characterizing the behavior of autonomous AI agents, including multi-agent interactions and emergent collaboration patterns.

Understand

Transparency & Interpretability

Techniques for making AI decision-making processes legible to humans — from mechanistic interpretability to explanation interfaces and audit tools.

Control

AI Control & AI Safety

Frameworks and mechanisms for maintaining meaningful human oversight of AI systems, including alignment verification, safety testing, and governance protocols.

Schedule

Final programme for both days.

Friday, June 5, 2026 Final

12:00–13:00	Arrival & Reception
13:00–13:20	COAI Welcome & Opening Remarks
13:20–13:30	Organizing Team Introduction
13:30–13:40	Message from IASEAI
13:40–14:20	Keynote 1 Vera Schmitt — XplaiNLP Group, TU Berlin Watch the recording of this talk
14:20–14:40	Paper Talk “Counterfactual Preference Perturbation Localises Objective Misalignment in Multi-Agent LLM Negotiation” Dominic Behling
14:40–15:00	Paper Talk “Architecture as Governance: Why LLM Agents Need Structural Compliance Enforcement” Steffen Höpfner, Sigurd Schacht Watch the recording of this talk
15:00–15:20	Paper Talk “Beyond Liars' Bench: The Impact of Lie Typology, Depth, and Sparsity on Deception Detection in LLMs” Amr Moustafa, Maximilian Feser, Florian Mai Watch the recording of this talk
15:30–16:00	Coffee & Snacks
16:00–16:10	Message from APART
16:10–16:25	Paper Talk “Agreement is not Alignment: Divergent Moral Grounds in Human and LLM Ethical Judgments” Octavian M. Machidon et al.
16:25–16:45	Paper Talk “From Recognition to Refusal: Mapping Safety Formation Zones Across Model Architectures” Carsten Lanquillon, Sigurd Schacht Watch the recording of this talk
16:45–18:00	Poster Session
from 19:00	Networking

Saturday, June 6, 2026 Final

09:00–10:20	Arrival · Coffee, Snacks, Networking
10:20–11:05	Keynote 2 Yonatan Belinkov — Harvard & Technion Watch the recording of this talk
11:05–11:50	Poster Session
11:50–12:10	Paper Talk “Organizational Integration of Autonomous, AI-Based Multi-Agent Systems: A Scenario Analysis” Sandra Kalogiannis
12:10–12:25	Group Photo
12:25–13:25	Lunch Break
13:25–14:10	Keynote 3 Joel Z. Leibo — Google DeepMind Watch the recording of this talk
14:10–14:55	Poster Session
14:55–15:15	Paper Talk “Rethinking Red Teaming: There Is More To Agents Than LLMs” Thomas Wacker, Carsten Lanquillon, Sigurd Schacht
15:15–15:35	Paper Talk “The Arbiter Agent: Continually Monitoring Multi-Agent Conversations to Detect Emergent Misalignment” Filippo Tonini et al. Watch the recording of this talk
15:35–16:05	Coffee & Snacks
16:05–16:25	Paper Talk “StreamFair: Online Fair Adaptation Under Temporal Intersectional Drift” Md Fahim Sikder, Ericka Johnson, Fredrik Heintz Watch the recording of this talk
16:25–16:45	Paper Talk “Confirming Process, Missing Function: The Uninstrumented Oversight Layer in AI Transparency and Control Research” Brian Moriarty
16:45–17:15	Closing Session

Talk Recordings

Keynotes and paper talks from AITC 2026, published with the speakers’ consent.

More talks may follow. We are still collecting publication consent from the remaining speakers, and recordings are released here as they are cleared — it is worth checking back at a later point. Speakers with questions about their own recording are welcome to get in touch.

As AI systems become more capable and autonomous, the need for transparency has never been greater. Understanding how these systems behave, interpret information, and can be controlled is essential for building AI that is safe and aligned with human values.

The AI Transparency Conference provides a dedicated space for researchers to share findings, challenge ideas, and shape the future of transparent AI — together.