Call for Proposal

Generative Artificial Intelligence Models (GAIMs) have emerged as transformative tools, driving advancements across diverse domains. However, their widespread adoption raises critical concerns across various dimensions, including truthfulness, safety, cybersecurity, fairness, robustness, and privacy. These models have demonstrated remarkable performance on knowledge reasoning tasks, owing to their implicit knowledge derived from extensive pretraining data. However, their inherent knowledge bases often suffer from disorganization and illusion, bias towards common entities, and rapid obsolescence. Consequently, GAIMs frequently make up untruthful information, exhibit resistance to updating outdated knowledge, or struggle with generalizing across multiple languages. The Capital One Illinois Center for Generative AI Safety, Knowledge Systems, and Cybersecurity (ASKS) is a collaboration between Capital One and the University of Illinois Urbana-Champaign, aiming to advance GAIMs’ up-to-date knowledge, safety, truthfulness, and trustworthiness across multiple dimensions and model types, and paving the way for safer and more responsible integration of GAIMs into critical applications including cybersecurity and situation understanding.

Research topic areas

The ASKS Center is calling for proposals for the second funding round, 2026-2027 academic year. In this phase, the center will support 3-5 sponsored projects. Each funded project will support a PhD research assistant and 2 weeks of summer salary for the principal investigators (PIs). Each selected project will then engage with research scientists from Capital One to collaborate and jointly advise PhD students.

Topics of interest would include, but are not limited to, those below. Please feel free to bring your unique viewpoint and expertise to these topics:

Large Language Models (LLMs) and Vision Language Models (VLMs) Safety:

•  LLM Reasoning
•  LLM Agents
•  AI Alignment
•  Synthetic Data Generation
•  LLM Security (e.g., Jailbreaking, Prompt Injection, Guardrail Models)
•  Hallucination Mitigation, LLM/VLM factuality
•  Multimodal Knowledge Representation
•  Knowledge Localization, Updating, and Editing
•  Unlearning (e.g., Debiasing through Unlearning Biases)
•  Structure-enhanced RAG (Retrieval Augmented Generation)
•  Privacy Preservation

Knowledge Systems:

• LLM-enhanced Information Extraction and Knowledge Graph Construction
• Situation Understanding, Forecasting, and Reporting

AI for CyberSecurity:

• Agentic AI for Novel Attacks and Defenses
• Multi-dimensional Anomaly Detection 
• LLM Explanation for Cybersecurity
• Temporal-spatial Aware Foundation Models for Cybersecurity
• Privacy-preserving Data Analytics

Call for Proposal Process

• December 11, 2025: Abstract Submission Deadline
• January 15, 2026: Abstract feedback sessions with Capital One scientists and UIUC Faculty
• February 16, 2026: Full Proposal Submission Due
• March 30, 2026: Full Proposal Acceptance Notification
• April 14, 2026: On Campus ASKS Center Event

Eligibility

• Eligibility: Full-time tenure-track, research-track and teaching faculty members at Illinois are eligible to submit proposals as PIs.