Projects and labs

Linguistically Motivated Language Model Security

Uses linguistic analysis to understand and mitigate security and privacy risks in multilingual language models.

• Security & privacy
• Multilingual NLP

PI

Johannes Bjerva

Funder

Novo Nordisk Foundation

Amount

DKK 9.9m

Building TRUST in Text

Linguistically Motivated Language Model Detection

Develops linguistically grounded ways to detect poisoned or tampered language models and scalable safeguards across languages and domains.

• Model integrity
• Linguistic fingerprints
• Explainable AI

PI

Johannes Bjerva

Funder

Independent Research Fund Denmark · Sapere Aude

Amount

DKK 6.19m

Partners

Stockholm University, NVIDIA

AI:PAGE-Lab

Turns historical texts into searchable, connected knowledge so researchers can discover patterns and relationships across cultural-heritage collections.

• Knowledge graphs
• Digital cultural heritage

PI

Russa Biswas and Johan Heinsen

Funder

Aalborg University · AI:X

Amount

ca. DKK 3.5m

Funding share

50% · Russa Biswas

AI:SECURITY

Connects NLP and cybersecurity to protect AI systems from hijacking and misuse and to counter AI-enabled disinformation and phishing.

• AI security
• Federated learning
• Disinformation

PI

Johannes Bjerva and Qiongxiu Li

Funder

Aalborg University · AI:X

Amount

DKK 7m

Funding share

50% · Johannes Bjerva

Mode Collapse Robust LLMs

Investigates how alignment training narrows the range of ideas expressed by language models, and develops ways to preserve that diversity while maintaining safety.

• Information diversity
• LLM alignment

PhD fellow

Luc Raszewski

Supervisors

Dustin Wright and Johannes Bjerva

Funder

Danish Advanced Research Academy · Novo Nordisk Foundation

Amount

ca. DKK 2m

Fast and Focused: Information-Structure-Aware AI for Accessibility

Studies how AI handles the way speakers signal familiar and new information, with the aim of improving the speed and reliability of assistive language technology.

• Information structure
• Accessible AI

PI

Marcell Richard Fekete

Funder

Independent Research Fund Denmark · International Postdoc

Amount

DKK 2,140,857

Partners

Budapest University of Technology and Economics

Formal Semantic Methods for AI Safety

Combines formal semantic analysis and neural interpretability to identify and explain rare multilingual failures in generative AI.

• Formal semantics
• AI safety
• Interpretability

PI

Johannes Bjerva

Funder

Coefficient Giving · Technical AI Safety Research

Amount

DKK 2.4m

Digital Twins for Abundant Feedback

Builds explainable multilingual models of expert assessment to support timely, useful feedback at scale.

• NLP for education
• Explainable feedback

PI

Johannes Bjerva and Euan Lindsay

Funder

Villum Foundation · Synergy

Amount

DKK 3m

Guarantees of Factuality in Financial KPI Extraction

Develops reliable methods and data for extracting financial KPIs from earnings transcripts with explicit factuality guarantees.

• Financial NLP
• Factuality
• Information extraction

PI

Johannes Bjerva

Funder

Innovation Fund Denmark · Industrial PhD

Amount

DKK 1.07m

Partners

Alipes ApS