Experts are warning about a surge in interest among cybercriminals for artificial intelligence tools on the dark web, seeking access to sensitive user data. Flare, a threat exposure management company, has identified over 200,000 OpenAI credentials for sale on the dark web. These credentials could potentially enable hackers to gain unauthorized access to company secrets and personal information.
The exposure of these credentials is attributed to info-stealing malware, which may contain even more information not intended for public viewing. While the number of at-risk credentials is relatively small compared to the vast number of ChatGPT users (estimated at 100 million), it is an increase from the approximately 101,000 credentials found in the logs of info-stealing malware earlier in June.
Furthermore, a malicious ChatGPT alternative, trained using data about malware, has been gaining popularity. Screenshots demonstrate how threat actors can use this AI chatbot to generate convincing-looking attacks that can be disseminated to victims via emails, advertisements, or web pages.
In response to previous cases of dark web-hosted credentials, OpenAI emphasized that it adheres to industry best practices for authenticating and authorizing users to its services. The company advises users to employ strong passwords and only install verified and trusted software on their personal computers.
It’s crucial to recognize that the exposure of credentials isn’t a result of flaws in OpenAI’s system, but rather due to info-stealing malware that could infiltrate systems through various entry points, such as fake ads and scam emails designed to plant malware.
To mitigate risks, Flare recommends regular dark web monitoring for individuals at risk, along with the use of up-to-date endpoint protection software that incorporates AI enhancements to improve threat detection. Additionally, companies are urged to practice good internet hygiene and regularly refresh staff training to stay vigilant against potential cyber threats.
