ESPERANTO: Evaluating Synthesized Phrases to Enhance Robustness in AI Detection for Text Origination
While large language models (LLMs) exhibit significant utility across various domains, they simultaneously are susceptible to exploitation for unethical purposes, including academic misconduct and dissemination of misinformation. Consequently, AI-generated text detection systems have emerged as a countermeasure. However, these detection mechanisms demonstrate vulnerability to evasion techniques and lack robustness against textual manipulations. This paper introduces back-translation as a novel technique for evading detection, underscoring the need to enhance the robustness of current detection systems. The proposed method involves translating AI-generated text through multiple languages before back-translating to English. We present a model that combines these back-translated texts to produce a manipulated version of the original AI-generated text. Our findings demonstrate that the manip- doi
- 10.1145/3720553.3746665
- isbn
- 979-8-4007-1534-1
- name
- ESPERANTO: Evaluating Synthesized Phrases to Enhance Robustness in AI Detection for Text Origination
- source
- supplied BITS XML and PDF
- acm_url
- https://dl.acm.org/doi/10.1145/3720553.3746665
- authors
- Navid Ayoobi, Lily Knab, Wen Cheng, David Pantoja, Hamidreza Alikhani, Sylvain Flamant, Jin Kim, Arjun Mukherjee
- doi_url
- https://doi.org/10.1145/3720553.3746665
- license
- © 2025 Copyright held by the owner/author(s). Publication rights licensed to ACM.
- summary
- While large language models (LLMs) exhibit significant utility across various domains, they simultaneously are susceptible to exploitation for unethical purposes, including academic misconduct and dissemination of misinformation. Consequently, AI-generated text detection systems have emerged as a countermeasure. However, these detection mechanisms demonstrate vulnerability to evasion techniques and lack robustness against textual manipulations. This paper introduces back-translation as a novel technique for evading detection, underscoring the need to enhance the robustness of current detection systems. The proposed method involves translating AI-generated text through multiple languages before back-translating to English. We present a model that combines these back-translated texts to produce a manipulated version of the original AI-generated text. Our findings demonstrate that the manip
- keywords
- (empty)
- arxiv_url
- https://arxiv.org/abs/2409.14285
- published
- 2025-09-15
- conference
- HT '25: 36th ACM Conference on Hypertext and Social Media, Chicago, IL, USA, September 15–19, 2025
- open_access
- false
- acm_html_url
- https://dl.acm.org/doi/full/10.1145/3720553.3746665
- ccs_concepts
- (empty)
- displayAuthor
- Navid Ayoobi, Lily Knab, Wen Cheng, David Pantoja, Hamidreza Alikhani, Sylvain Flamant, Jin Kim, Arjun Mukherjee
- proceedings_url
- https://dl.acm.org/doi/proceedings/10.1145/3720553
- displayPublishTime
- 2025-09-15
- acm_reference_format
- (empty)
Powered by Seed HypermediaOpen App