Text Adversarial Attacks With Dynamic Outputs

Summary

Researchers introduce the Textual Dynamic Outputs Attack (TDOA), a text adversarial attack for settings where the number and content of model labels vary, such as LLM outputs and multi-label classification. TDOA trains a clustering-based surrogate model that approximates dynamic fine-grained outputs with static coarse-grained labels, and uses a farthest-label targeted strategy to induce larger output changes. Evaluated on five datasets and ten victim models including GPT-4o and GPT-4.1, it reaches an 80.8% maximum attack success rate with five queries per text.

Topics

• Adversarial machine learning

Related items

• InfoPerplexity launches its own take on AI agents for workSame vendor · The Verge (AI)
• MediumBinance heist is stark example of AI governance gaps and basic security failuresSame vendor · CSO Online
• MediumFake OpenAI, SpaceX events used to steal Microsoft, Google accountsSame vendor · BleepingComputer
• LowOpenAI’s GPT-6 Astra silently obstructed safety research tasksSame vendor · BleepingComputer
• InfoWinner of the Prix Goncourt literary prize admits to using AISame vendor · The Guardian Technology