South Korean artificial intelligence firm AI-NOMIS had 47.8 GB of data with 93,485 explicit AI-generated images depicting underage individuals and JSON files belonging to its AI image generator platform GenNomis inadvertently exposed by a misconfigured database, Hackread reports.
Also included in the leaked data trove were face-swapped images and various X-rated deepfakes depicting Michelle Obama, Ariana Grande, and other celebrities as children, according to an investigation by cybersecurity researcher Jeremy Fowler published on vpnMentor. Additional details regarding GenNomis' processes were also revealed by the JSON files with command prompt logs and generated image links. Such a data leak is indicative of the potential exploitation of AI image generation, which should prompt increased developers to implement additional safeguards, including explicit deepfake detection systems, said Fowler, who warned of the possible utilization of the malicious images in extortion and revenge attacks. AI-NOMIS has already removed the database after the takedown of its website.
An In-Depth Guide to AI
Get essential knowledge and practical strategies to use AI to better your security program.
HalluSquatting, short for adversarial hallucination squatting, leverages the inability of LLMs to accurately distinguish between legitimate and fabricated resource locations, according to a paper by researchers from Tel Aviv University, Technion and Intuit.
The NCSC's Cyber Shield initiative is a response to the escalating scale, speed, and sophistication of cyber threats, particularly from state-sponsored actors and organized crime targeting public services.
Radware's update provides improved monitoring and mapping of agent interactions and dependencies, allowing security teams to understand how autonomous systems connect to enterprise resources.
Get daily email updates
SC Media's daily must-read of the most current and pressing daily news
