Training Description
Key Takeaways
- Focuses on human-centered privacy and security for language models
- Bridges technical mechanisms with human needs and policy
- Targets researchers and practitioners in AI, HCI, security, privacy, and governance
- Addresses real-world risks, user control, and accountability in LM systems
- Emphasizes interdisciplinary collaboration and best practices
The 2nd Workshop on Human-Centered Privacy and Security for Language Models (HAIPS@COLM 2026) will take place on October 9, 2026, in San Francisco. This event is co-located with COLM 2026 and is dedicated to advancing the understanding and development of privacy and security protections for language models as they become integral to society.
Event Overview
HAIPS@COLM 2026 brings together experts from diverse fields to examine the privacy and security challenges posed by language models. The workshop is designed to foster interdisciplinary dialogue, connecting the language modeling, human-computer interaction, usable security, and policy communities. The primary aim is to ensure that language model systems are not only technically robust but also aligned with human values and societal needs.
Participants will engage in discussions and presentations that critique current practices and propose new approaches for safeguarding user data and ensuring responsible deployment of language models. The event encourages collaboration to bridge gaps between technical solutions and the expectations of users and stakeholders.
Main Topics and Themes
The workshop covers a broad range of topics, including human expectations, consent, and control over data in language model systems. It explores privacy- and security-preserving infrastructure, user-facing protections, and risk prevention strategies. Additional themes include human oversight, accountability in agentic systems, and the evaluation of privacy, security, and safety in real-world deployments.
Other areas of focus involve the human impacts of adversarial threats, user vulnerability, manipulation, and the use of language models as simulated users or evaluators. The event also addresses organizational responsibilities and the development of human-centered policy and governance frameworks for language model privacy and security.
Audience and Experience
This workshop is tailored for researchers and practitioners in language modeling, artificial intelligence, human-computer interaction, security, privacy, and policy. Attendees include academics, industry professionals, policy analysts, and those involved in AI deployment and governance. The event offers a platform for education, thought leadership, and community building, with a strong emphasis on research exchange and interdisciplinary networking.
HAIPS@COLM 2026 features invited speakers, paper presentations, and recognition for outstanding contributions. The in-person format, co-located with a major conference, provides a dynamic environment for sharing insights and advancing best practices in human-centered AI security and privacy.