Independent Applied Research · Zürich

Independent research on how AI systems behave after deployment.

AI Resilience Lab studies how advanced AI systems fail, drift and exceed intended authority once they operate with tools, data, permissions and increasing autonomy.

The research question

How do we ensure AI systems remain accountable to people as they gain more autonomy, access and decision-making authority?

02 What we investigate

Five questions define the current research agenda.

01

Agentic AI and autonomous behaviour

How do agents behave once they can plan, call tools and act across systems?

02

Authority, identity and permissions

What happens when AI systems act through human or shared credentials?

03

Post-deployment AI risk

How do systems drift or produce unintended outcomes in real environments?

04

Monitoring and behavioural governance

What evidence shows an action was not only authorised, but intended?

05

Accountability and operational governance

Who answers when AI systems perform consequential actions?

03 How we research

AI governance analysed only through regulation describes what should happen. Four perspectives keep the findings tied to what systems do.

01

Behaviour

What AI systems actually do in operational environments.

02

Controls

Permissions, identity, monitoring, escalation and technical safeguards.

03

Governance

Accountability structures, decision rights and regulatory expectations.

04

Evidence

Incidents, experiments and documented cases turned into findings.

04 Inside the Lab

Research in progress, open questions and observations — clearly distinguished from published findings.

Open question

What does an audit trail have to record before an agent's action can be called intended?

Transaction logging answers whether a step was authorised. It does not answer whether the sequence of steps served the objective the agent was given.

Open question

Where does accumulated agent memory stop being context and start being an ungoverned control surface?

Memory persists across sessions and shapes later behaviour, while most governance boundaries are drawn around models and prompts.

05 From research to practice

Research that stays in a paper does not change how a system behaves on a Tuesday afternoon. Findings from the lab are translated into control models, assessment approaches, monitoring structures and operating principles that survive contact with a production environment — carried by four named frameworks: the Behavioural Governance Framework, the AI Agent Governance Stack, the AI Control Failure Taxonomy and the risk category AI Agent Fraud.

Applied work tests whether those methods hold under real operational constraints, and helps sharpen the questions the lab investigates.

Read the frameworks →

Research collaboration

Joint investigations, applied research and expert exchange.

Applied advisory

Focused work where AIRL research supports concrete governance, resilience or risk decisions.

Speaking & expert contribution

Panels, executive briefings, workshops and expert discussions.

Discuss a collaboration

A short conversation establishes the question at hand and whether the lab's current research is relevant to it. Based in Zürich, working across Switzerland and the EU, in English and German.

aljona@airesiliencelab.com