University of Maryland
UC Berkeley
Juan Felipe Cerón UribeAI Alignment Research Engineer · OpenAIResearcher↗
Researcher
Eric Wallace
OpenAIMember of Technical Staff
Wallace exposed language-model weaknesses through universal adversarial triggers—short phrases that derail responses.[1] He co-developed instruction hierarchy, teaching AI to prioritize trusted instructions over malicious inputs.[2] His safety, privacy and alignment work has contributed to GPT-5, deep research and ChatGPT’s agent mode.[3]
Career & education
Current roleOpenAI2023-presentMember of Technical Staff
OpenAI2023-presentPreviously
Google DeepMind2023-2023Research Intern
Meta2021-2021Research Intern
Google DeepMind2023-2023Meta2021-2021Education
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.
citations41,955As of Nov 6, '26
citations+7,660Since last snapshot, Jun 15, '26
References
- Universal Adversarial Triggers for Attacking and Analyzing NLParxiv.org
First author; short input-independent phrases expose systematic vulnerabilities.
- The Instruction Hierarchyarxiv.org
Equal first author; training models to respect differing trust levels to resist malicious instructions.
- Eric Wallace — current researchericswallace.com
Current personal site explicitly describes research on safety, robustness, privacy and alignment and contributions to named core models/products.
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 August 8, 2026 · Checked for accuracy and new information.
