Scientists at the Observatory on Social Media at Indiana University recently conducted a study uncovering a bot problem on X (formerly known as Twitter). The study revealed that there are approximately 1,140 AI-powered accounts on X that generate machine-generated content and steal selfies to create fake personas. This network of fake accounts, known as the “Fox8” botnet, uses ChatGPT to generate content with the intention of promoting suspicious websites and spreading harmful content.
The researchers found that these bot accounts not only attempt to convince people to invest in fake cryptocurrencies but are also believed to engage in stealing from existing crypto wallets. They frequently use hashtags such as #bitcoin, #crypto, and #web3 and engage with human-run accounts like Forbes’ crypto-centered X account (@ForbesCrypto) and the blockchain-centered news site Watcher Guru (@WatcherGuru). The impact of these bot accounts goes beyond looting crypto, as they distort online conversations and spread misinformation in various contexts, including elections and public health crises.
To appear more human-like, the Fox8 botnet’s harmful and spam accounts go beyond using stolen photos from real users. They also frequently interact with each other through retweets and replies, possess profile descriptions, and have a certain number of followers, friends, and tweets on average. According to the researchers, these elements suggest that these bots are actively participating in activities on X, making them more believable to human users.
The profiles of the Fox8 bots commonly mention cryptocurrencies and blockchains, with most of them being created over seven years ago, while some were created in 2023. In the past, botnets like Fox8 were easily noticeable due to their unconvincing content and tweets with unnatural language. However, advancements in language models, particularly ChatGPT, have made it increasingly difficult to detect accounts within the Fox8 botnet. These accounts spam human users with AI-generated posts in an attempt to convince them to invest in fake cryptocurrencies and are even suspected of stealing from existing crypto wallets.
The researchers express concern about the exploitation of free AI APIs like ChatGPT to deceive people. They found that even when using a large language model (LLM) content detector, the latest technology was unable to effectively distinguish between human and LLM-powered bots. The researchers did not disclose the specific handles associated with the Fox8 botnet accounts but managed to identify them through self-revealing tweets posted accidentally. The researchers concluded that 76% of the identified accounts are likely humans posting or retweeting ChatGPT outputs, while the remaining accounts are likely bots using LLMs for content generation.
The study’s main findings highlight the need for significant resources to develop appropriate countermeasures and regulations since there are currently no effective methods to detect AI-generated content. The researchers suggest that this study is just the tip of the iceberg and that more careful bad actors could develop malicious bots that are undetectable. They emphasize the importance of addressing these issues to protect users from the potentially harmful effects of bot activity on social media platforms like X.
Upon publication of the study, X reportedly took down the 1,140 illegitimate bots. However, the researchers did not engage with X regarding their findings, considering the platform to be unresponsive. The Post has reached out to X for comment and is awaiting a response.
Denial of responsibility! VigourTimes is an automatic aggregator of Global media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, and all materials to their authors. For any complaint, please reach us at – [email protected]. We will take necessary action within 24 hours.
