Exploring the Deceptive Power of LLM-Generated Fake News: A Study of Real-World Detection Challenges (2403.18249v2)
Abstract: Recent advancements in LLMs have enabled the creation of fake news, particularly in complex fields like healthcare. Studies highlight the gap in the deceptive power of LLM-generated fake news with and without human assistance, yet the potential of prompting techniques has not been fully explored. Thus, this work aims to determine whether prompting strategies can effectively narrow this gap. Current LLM-based fake news attacks require human intervention for information gathering and often miss details and fail to maintain context consistency. Therefore, to better understand threat tactics, we propose a strong fake news attack method called conditional Variational-autoencoder-Like Prompt (VLPrompt). Unlike current methods, VLPrompt eliminates the need for additional data collection while maintaining contextual coherence and preserving the intricacies of the original text. To propel future research on detecting VLPrompt attacks, we created a new dataset named VLPrompt fake news (VLPFN) containing real and fake texts. Our experiments, including various detection methods and novel human study metrics, were conducted to assess their performance on our dataset, yielding numerous findings.
- AFPFactCheck. Afpfactcheck. https://factcheck.afp.com/, 2023.
- How effectively can machines defend against machine-generated fake news? an empirical study. In Proceedings of the First Workshop on Insights from Negative Results in NLP, pp. 48–53, 2020.
- Adversarial attacks and defences: A survey. arXiv preprint arXiv:1810.00069, 2018.
- CheckYourFact. Checkyourfact. https://checkyourfact.com/, 2023.
- Can llm-generated misinformation be detected? arXiv preprint arXiv:2309.13788, 2023.
- Vicuna: An open-source chatbot impressing gpt-4 with 90%* chatgpt quality. See https://vicuna. lmsys. org (accessed 14 April 2023), 2023.
- BERT: pre-training of deep bidirectional transformers for language understanding. CoRR, abs/1810.04805, 2018. URL http://arxiv.org/abs/1810.04805.
- FactCheck. Factcheck. https://www.factcheck.org/, 2023.
- Making pre-trained language models better few-shot learners. arXiv preprint arXiv:2012.15723, 2020.
- Walid Hariri. Unlocking the potential of chatgpt: A comprehensive exploration of its applications, advantages, limitations, and future directions in natural language processing. arXiv preprint arXiv:2304.02017, 2023.
- HealthFeedback. Healthfeedback. https://healthfeedback.org/, 2023.
- Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685, 2021.
- Disinformation detection: An evolving challenge in the age of llms. arXiv preprint arXiv:2309.15847, 2023.
- LeadStories. Leadstories. https://leadstories.com/, 2023.
- Jailbreaking chatgpt via prompt engineering: An empirical study. arXiv preprint arXiv:2305.13860, 2023.
- Roberta: a robustly optimized bert pretraining approach (2019). arXiv preprint arXiv:1907.11692, 364, 2019.
- Contrasting linguistic patterns in human and llm-generated text. arXiv preprint arXiv:2308.09067, 2023.
- NIH. Nih. https://www.nih.gov/, 2023.
- University of Oregon Library. Fake news and information literacy. https://researchguides.uoregon.edu/fakenews/issues/defining.
- OpenAI. Chatgpt 3.5. https://chat.openai.com/chat, 2023.
- On the risk of misinformation pollution with large language models. arXiv preprint arXiv:2305.13661, 2023.
- Red teaming language models with language models. arXiv preprint arXiv:2202.03286, 2022.
- PolitiFact. Politifact. https://www.politifact.com/, 2023.
- Declare: Debunking fake news and false claims using evidence-aware deep learning. arXiv preprint arXiv:1809.06416, 2018.
- ” do anything now”: Characterizing and evaluating in-the-wild jailbreak prompts on large language models. arXiv preprint arXiv:2308.03825, 2023.
- defend: Explainable fake news detection. In Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining, pp. 395–405, 2019.
- Adapting fake news detection to the era of large language models. arXiv preprint arXiv:2311.04917, 2023.
- Exploring adversarial examples in malware detection. In 2019 IEEE Security and Privacy Workshops (SPW), pp. 8–14. IEEE, 2019.
- Med-mmhl: A multi-modal dataset for detecting human-and llm-generated misinformation in the medical domain. arXiv preprint arXiv:2306.08871, 2023.
- Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288, 2023.
- ungjus. Fn-bert. https://huggingface.co/ungjus/Fake_News_BERT_Classifier, 2023.
- Disinformation capabilities of large language models. arXiv preprint arXiv:2311.08838, 2023.
- Implementing bert and fine-tuned roberta to detect ai generated news by chatgpt. arXiv preprint arXiv:2306.07401, 2023.
- WebMD. Webmd. https://www.webmd.com/, 2023.
- Chain-of-thought prompting elicits reasoning in large language models. Advances in Neural Information Processing Systems, 35:24824–24837, 2022.
- Fake news in sheep’s clothing: Robust fake news detection against llm-empowered style attacks. arXiv preprint arXiv:2310.10830, 2023.
- Defending against neural fake news. Advances in neural information processing systems, 32, 2019.
- Mining dual emotion for fake news detection. In Proceedings of the web conference 2021, pp. 3465–3476, 2021.
- A survey of fake news: Fundamental theories, detection methods, and opportunities. ACM Computing Surveys (CSUR), 53(5):1–40, 2020.
- Yanshen Sun (8 papers)
- Jianfeng He (32 papers)
- Limeng Cui (19 papers)
- Shuo Lei (10 papers)
- Chang-Tien Lu (54 papers)