More than 100,000 user accounts of the popular artificial intelligence chatbot platform ChatGPT have been compromised over the last year using information-stealing malware, a new report has revealed. The report, published by Singapore-based cybersecurity firm Group-IB, identified 101,134 compromised accounts, the credentials of many of which have been traded over the last year on illicit dark web marketplaces. At its peak in May, nearly 27,000 credentials of compromised ChatGPT accounts were traded on the dark web, the group noted, adding that the Asia-Pacific region experienced the highest concentration of ChatGPT credentials offered for sale. This region, according to the report, accounted for almost 40 per cent of compromised accounts between June 2022 and May 2023, followed by Europe. Since its widespread rollout in November last year, ChatGPT has seen growing use, with employees taking advantage of the chatbot to optimise their work across fields from software development to business communications. As the chatbot stores the history of user queries and the AI’s responses, experts have warned that unauthorised access to ChatGPT accounts could expose confidential or sensitive information. “Employees enter classified correspondences or use the bot to optimize proprietary code. Given that ChatGPT’s standard configuration retains all conversations, this could inadvertently offer a trove of sensitive intelligence to threat actors if they obtain account credentials,” said Dmitry Shestakov, the head of threat intelligence at Group-IB. Several businesses, institutions and universities across the world, including several in Japan, have either banned use of the chatbot, or have warned staff to not reveal sensitive information to the AI bot as such data can be exploited for targeted attacks against companies and their employees. The Singapore-based cybersecurity group warned in its latest report that ChatGPT accounts have already gained popularity within underground communities on the dark web that are accessible only via special software. Using malicious software known as info stealers, credentials saved in browsers, bank card details, crypto wallet information, cookies, browsing history and other information from browsers installed on infected computers are being stolen and sent to operators. Logs containing user information, including data on the IP addresses, are being actively traded on dark web marketplaces, according to Group-IB. A majority of logs containing ChatGPT accounts have been breached by the infamous Raccoon info stealer, the group noted. Experts urge users to update passwords regularly and implement two-factor authentication for accessing their ChatGPT accounts. Users are also advised to disable the chatbot’s chat saving feature from its settings menu or manually delete conversations immediately after use. Read More ChatGPT ‘grandma exploit’ gives users free keys for Windows 11 Protect personal data when introducing AI, privacy watchdog warns businesses How Europe is leading the world in the push to regulate AI Scientists warn of threat to internet from AI-trained AIs ChatGPT ‘grandma exploit’ helps people pirate software Hundreds attend ‘soulless’ AI-generated church service