Papers by Julia Soares Dollis
Safety Is Not Universal: The Selective Safety Trap in LLM Alignment (2026.findings-acl)
Copied to clipboard
Iago Alves Brito, Walcy Rios, Julia Soares Dollis, Diogo Fernandes Costa Silva, Arlindo Rodrigues Galvão Filho
| Challenge: | Existing safety evaluations of large language models aggregate harms under generic categories such as "Identity Hate" a bilingual benchmark identifies a selective safety trap, where defense rates vary by up to 42% within the same model solely based on the target group. |
| Approach: | They propose a bilingual adversarial benchmark to audit selective safety in large language models . defense rates vary by up to 42% within the same model solely based on target group . |
| Outcome: | The proposed benchmark identifies a selective safety trap in large language models . defense rates vary by up to 42% within the same model solely based on the target group. |
Modeling, Evaluating, and Embodying Personality in LLMs: A Survey (2025.findings-emnlp)
Copied to clipboard
Iago Alves Brito, Julia Soares Dollis, Fernanda Bufon Färber, Pedro Schindler Freire Brasil Ribeiro, Rafael Teixeira Sousa, Arlindo Rodrigues Galvão Filho
| Challenge: | This survey provides a comprehensive overview of the LLM-driven personality scenario. |
| Approach: | This survey provides a comprehensive overview of the LLM-driven personality scenario. |
| Outcome: | The proposed taxonomy analyzes the limitations of existing methods and identifies key research gaps. |
Proxy Barrier: A Hidden Repeater Layer Defense Against System Prompt Leakage and Jailbreaking (2025.findings-emnlp)
Copied to clipboard
Pedro Schindler Freire Brasil Ribeiro, Iago Alves Brito, Rafael Teixeira Sousa, Fernanda Bufon Färber, Julia Soares Dollis, Arlindo Rodrigues Galvão Filho
| Challenge: | Prompt injection and jailbreak attacks remain a critical vulnerability for large language models . a lightweight defense that interposes a proxy LLM between the user and the target model addresses this vulnerability . |
| Approach: | a lightweight proxy LLM is interposed between the user and the target model to prevent prompt injection and jailbreak attacks. |
| Outcome: | ProB outperforms baselines and achieves up to 98.8% defense effectiveness . it is deployable entirely at the API level and requires no access to model weights or prompts . |