English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Open

Large Language Models (LLMs) have achieved human-level fluency in textgeneration, making it difficult to distinguish between human-written andLLM-generated texts. This poses a growing risk of misuse of LLMs and demandsthe development of detectors to identify LLM-generated texts. However, existingdetectors lack robustness against attacks: they degrade detection accuracy bysimply paraphrasing LLM-generated texts. Furthermore, a malicious user mightattempt to deliberately evade the detectors based on detection results, butthis has not been assumed in previous studies. In this paper, we proposeOUTFOX, a framework that improves the robustness of LLM-generated-textdetectors by allowing both the detector and the attacker to consider eachother's output. In this framework, the attacker uses the detector's predictionlabels as examples for in-context learning and adversarially generates essaysthat are harder to detect, while the detector uses the adversarially generatedessays as examples for in-context learning to learn to detect essays from astrong attacker. Experiments in the domain of student essays show that theproposed detector improves the detection performance on the attacker-generatedtexts by up to +41.3 points F1-score. Furthermore, the proposed detector showsa state-of-the-art detection performance: up to 96.9 points F1-score, beatingexisting detectors on non-attacked texts. Finally, the proposed attackerdrastically degrades the performance of detectors by up to -57.0 pointsF1-score, massively outperforming the baseline paraphrasing method for evadingdetection.

OUTFOX: LLM-Generated Essay Detection Through In-Context Learning with Adversarially Generated Examples