The US Cybersecurity and Infrastructure Security Agency (CISA) has issued a new binding operational directive, AI-25-01, mandating AI model provenance reporting for all federal agencies and their crit
AIBW Security DeskA new research paper published by the Stanford AI Lab details a novel jailbreak technique named 'Semantic Splicing'. This method is designed to circumvent the sophisticated safety and alignment filter
AIBW Security DeskIn a joint directive, the U.S. Department of Commerce and the European Commission have announced the 'Trans-Atlantic AI Secure Development Framework' (TAISDF). This new regulation mandates that develo
AIBW Security DeskHealthMind Inc., a prominent AI-driven diagnostic service, has disclosed a critical data breach affecting approximately 5 million patients. Attackers leveraged a sophisticated model inversion attack t
AIBW Security DeskResearchers from Carnegie Mellon University have published a paper detailing a novel jailbreak technique named 'Sleepwalker'. This method utilizes a combination of steganographically encoded instructi
AIBW Security DeskMicrosoft has released 'Guardian,' a new open-source framework designed to help organizations secure their end-to-end AI development and deployment pipelines. Guardian integrates with popular MLOps pl
AIBW Security DeskSynthCorp, a leading provider of AI-powered voice synthesis services, has disclosed a major data breach affecting over two million users. The breach resulted from a misconfigured cloud storage bucket
AIBW Security DeskResearchers at the Stanford AI Lab have published a paper detailing a novel jailbreak technique called 'Semantic Splicing.' This method bypasses safety filters in leading Large Language Models by embe
AIBW Security DeskThe National Institute of Standards and Technology (NIST) has officially released the AI Integrity and Defense Auditing (AIDA) framework, a comprehensive open-source toolkit for security professionals
AIBW Security DeskA team from Stanford's AI Lab has published a groundbreaking paper detailing a new jailbreak technique named 'Cognitive Override.' This method exploits the way large multi-modal models (LMMs) process
AIBW Security DeskA major data breach at SynthVerse AI, a leading provider of enterprise AI agents, has exposed over 50 terabytes of sensitive data. The breach, attributed to a misconfigured cloud storage bucket, inclu
AIBW Security DeskA new paper published by researchers at Carnegie Mellon University's CyLab introduces a novel jailbreak technique named 'Semantic Splicing'. This method bypasses the safety filters of leading large la
AIBW Security Desk