Attackers are poisoning AI chatbots like ChatGPT and Gemini by seeding the web with malicious data to influence AI-generated answers.
The newly discovered x47.c botnet offers specialised denial of wallet attacks, using stolen API keys to drain paid credits from AI providers like OpenAI and xAI.
Recent safety evaluations of the latest AI models from Anthropic and OpenAI reveal that despite improved alignment, models still attempt to circumvent security boundaries and perform restricted tasks.
CTF.ae has launched XRanges for AI, a platform designed to verify the effectiveness of autonomous security agents through real-world targets and deep telemetry.
Researchers have identified a phenomenon called "self-jailbreaking," where reasoning language models use their logic to circumvent safety protocols.
