ACL Findings2025
When Detection Fails: The Power of Fine-Tuned Models to Generate Human-Like Social Media Text
Abstract
Detecting AI-generated text is a difficult problem to begin with; detecting AI-generated text on social media is made even more difficult due to the short text length and informal, idiosyncratic language of the internet. It is nonetheless important to tackle this problem, as social media represents a significant attack vector in online influence campaigns, which may be bolstered through the use of mass-produced AIgenerated posts supporting (or opposing) particular policies, decisions, or events. We approach this problem with the mindset and resources of a reasonably sophisticated threat actor, and create a dataset of 505,159 AI-generated social media posts from a combination of opensource, closed-source, and fine-tuned LLMs, covering 11 different controversial topics. We show that while the posts can be detected under typical "research" assumptions about knowledge of and access to the generating models, under the more realistic assumption that an attacker will not release their fine-tuned model to the public, detectability drops dramatically. This result is confirmed with a human study. Ablation experiments highlight the vulnerability of various detection algorithms to fine-tuned LLMs. This result has implications across all detection domains, since fine-tuning is a generally applicable and realistic LLM use case.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- QLoRA: Efficient Finetuning of Quantized LLMs cites LoRA: Low-Rank Adaptation of Large Language Models
- Paraphrasing evades detectors of AI-generated text, but retrieval is an effective defense cites DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvature
- DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvature cites Paraphrasing evades detectors of AI-generated text, but retrieval is an effective defense
- Spotting LLMs With Binoculars: Zero-Shot Detection of Machine-Generated Text cites Paraphrasing evades detectors of AI-generated text, but retrieval is an effective defense
- Spotting LLMs With Binoculars: Zero-Shot Detection of Machine-Generated Text cites DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvature
Builds on13
- LoRA: Low-Rank Adaptation of Large Language ModelsEdward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu et al.ICLR 2022 · 18,833 citations
- QLoRA: Efficient Finetuning of Quantized LLMsTim Dettmers, Artidoro Pagnoni, Ari Holtzman, Luke ZettlemoyerNeurIPS 2023 · 5,863 citations
- DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability CurvatureEric Mitchell, Yoonho Lee, Alexander Khazatsky, Christopher D. Manning et al.ICML 2023 · 988 citations
- Paraphrasing evades detectors of AI-generated text, but retrieval is an effective defenseKalpesh Krishna, Yixiao Song, Marzena Karpinska, John Wieting et al.NeurIPS 2023 · 657 citations
- Spotting LLMs With Binoculars: Zero-Shot Detection of Machine-Generated TextAbhimanyu Hans, Avi Schwarzschild, Valeriia Cherepanova, Hamid Kazemi et al.ICML 2024 · 262 citations
Related papers
- MultiSocial: Multilingual Benchmark of Machine-Generated Text Detection of Social-Media TextsDominik Macko, Jakub Kopal, Róbert Móro, Ivan SrbaACL 2025 · 15 citations
- Fingerprinting Fine-tuned Language Models in the WildNirav Diwan, Tanmoy Chakraborty, Zubair ShafiqACL Findings 2021 · 3 citations
- Is Your Paper Being Reviewed by an LLM? Benchmarking AI Text Detection in Peer ReviewSungduk Yu, Man Luo, Avinash Madasu, Vasudev Lal et al.ICLR 2026 · 24 citations
- MAGE: Machine-generated Text Detection in the WildYafu Li, Qintong Li, Leyang Cui, Wei Bi et al.ACL 2024 · 44 citations
- XDAC: XAI-Driven Detection and Attribution of LLM-Generated News Comments in KoreanWooyoung Go, Hyoungshick Kim, Alice Oh, Yongdae KimACL 2025
