Check out our research summary for our news environment perceived fake news detection in AI Ethics Brief. Montreal AI Ethics Institute montrealethics.ai/zoom-out-and-o…

We have released a curated list of AI/machine-generated text and will update it constantly! LLMs, detection & attribution methods, products & services, datasets... all you want to know about the machine-generated text is here~ #AIGC #AIgeneratedtext github.com/ICTMCG/Awesome…

Our book chatper on the role of visual information in fake news detection has obtained its 100th citation! scholar.google.com/citations?view…

LLMs can help with fake news detection, but how in reality? We reveal that being an ADVISOR for fine-tuned small models is a better role than being an ACTOR for LLMs like GPT-3.5. See our #AAAI2024 paper: arxiv.org/pdf/2309.12247

Internal states of LLMs are useful for AI-generated text detection/attribution but how to do it for *black-box* LLMs? We design an efficient re-sampling method to estimate black-box LLMs' output prob. Only 10 selected words lead to good performance! See arxiv.org/pdf/2402.09199