Kamile Lukosiute Joins OpenAI Foundation as Program Officer

Kamile Lukosiute has joined the OpenAI Foundation as a Program Officer, where she will build the organization’s cyber resilience workstream. Joining the Foundation, she brings experience in AI security, cyber misuse research, frontier model evaluations, and technology risk.

Kamile previously worked as a Research Scholar at GovAI, where she spent a year studying the cybersecurity risks associated with increasingly capable AI systems. Her research examined AI-enabled cyber misuse, extreme misuse scenarios, model evaluations, capability thresholds, and safeguards for frontier AI systems.

Key Highlights

• Kamile Lukosiute joins the Foundation as Program Officer
• She will build its new cyber resilience workstream
• She brings expertise in AI security and cyber risks

Before joining GovAI, Kamile developed her expertise through roles in AI security at Cisco and Robust Intelligence, as well as a residency at Anthropic between 2022 and 2023. Robust Intelligence was acquired by Cisco in October 2024, adding to her experience across AI security research and enterprise technology.

Her recent research has focused on understanding the scale and implications of AI-related cyber risks. Earlier this year, Kamile co-authored research with John Halstead and Luca Righetti that estimated the global economic cost of cybercrime at approximately $500 billion annually. The research combined 27 existing estimates with evidence covering individual and business losses, response costs, and cybersecurity spending.

The researchers placed the estimate within a 90 percent confidence interval of $100 billion to $1 trillion, creating a baseline for evaluating how AI could affect the future scale of cybercrime and when additional safeguards may be required.

Kamile has also examined the growing capabilities of open-weight AI models and their implications for  cybersecurity. In February, she argued that open-weight models were approximately three to six months behind leading closed models on available capability measures and could soon become capable of supporting offensive cyber operations.

Her research has increasingly emphasized cyber resilience, AI safety, women in technology, responsible AI, and emerging technology risk, particularly the need to strengthen defenses rather than depend solely on restricting access to advanced models.

“We need to build the next generation of controls and deploy them to the infrastructure we all depend on,” Kamile wrote, pointing to payments, energy, water, and medicine as critical systems increasingly dependent on computing infrastructure.

She has proposed strengthening protections for potential targets, developing AI-powered defensive tools suited to operational environments, and creating detection capabilities that do not depend entirely on AI developers knowing how their models are being used.

Kamile has also clarified that the OpenAI Foundation is a separate entity from OpenAI, the company that develops AI models. “The Foundation is incredibly well-placed to support this work,” she said, describing its purpose as ensuring that AI benefits humanity.

Her published work has continued to explore how AI security should be assessed in real-world environments. A January 2025 paper co-authored with Adam Swanda argued that existing LLM cyber evaluations may overlook operational risks by placing greater emphasis on model performance than on threat-actor behavior and real-world consequences.

With her latest appointment, Lukošiūtė brings together her research in AI security, cyber risk, responsible AI, and frontier model safety to develop the Foundation’s cyber resilience workstream. She will work alongside Lennart Heim, Yonadav Shavit, Wojciech Zaremba, and others as the initiative takes shape.