Generative AI

The “Deceptive Delight” method tricks models into generating harmful content within two or three interactions.

Okta's Shiven Ramji dives into the importance of protecting the identity of AI agents and the issues consumers and businesses have with generative AI.

More than 20 adverse operations were interrupted by OpenAI in 2024, a new report revealed.

Successful LLM attacks result in sensitive data leakage 90% of the time, a Pillar Security study found.

The many lessons to take away from a 24-year old flaw in glibc and the mastery in crafting an exploit in PHP, changing a fuzzer's configuration to find more flaws, fuzzing LLMs for prompt injection and jailbreaks, security hardening of baseband code, revisiting the threat models in Microsoft's Recall, and more!

Zed Attack Proxy has been a crucial web app testing tool for decades. It's also had a struggle throughout 2024 to obtain funding that would enable the tool to add more features while remaining true to its open source history. Simon Bennetts, founder of ZAP, and Ori Bendet from Checkmarx update us on that journey, share some exploration of LLM fuzzi...

Phishing attacks are become more prevalent as threat actors are arming themselves with potent new tools.