UCLA
University of Patras
Nicholas CarliniMember of Technical Staff · AnthropicResearcher↗
Researcher
Andreas Terzis
AnthropicMember of Technical Staff and Manager, Safeguards
Andreas Terzis develops defenses for AI systems. He co-authored research testing whether hidden instructions could hijack Gemini[1] and helped develop a safeguard that lets Claude flag its own potentially harmful responses.[2] He has also worked on creating training data while protecting users’ privacy.[3]
Career & education
Current roleAnthropic2026-presentMember of Technical Staff and Manager, Safeguards
Anthropic2026-presentPreviously
Google DeepMind2023-2025Senior Staff Software Engineer
Google2013-2020Software Engineer
Johns Hopkins University2010-2012Associate Professor
Google DeepMind2023-2025Google2013-2020Johns Hopkins University2010-2012Education
Work and education history is primarily focused on AI-relevant roles and may not be comprehensive.
Citation Trend
Total citations to scholarly work, including research papers, technical reports, books and theses, as recorded by Google Scholar. Each point is a dated snapshot.
citations27,569As of Nov 13, '26
References
- Lessons from Defending Gemini Against Indirect Prompt Injectionsarxiv.org
Co-authored report on testing Gemini against hidden instructions in external content.
- Scaling Misuse Detection with Self-Assessmentsafeguards.anthropic.com
Co-authored the Anthropic report on a safeguard deployed in Claude.
- Protecting users with differentially private synthetic training dataresearch.google
Named collaborator on privacy-preserving synthetic training data.
Based on public sources. If accuracy is important, verify details using the links or references.
For corrections, updates, or requests to remove information, contact us.
Reviewed November 13, 2026 · Checked for accuracy and new information.
