Agentic AI delivers autonomous operations based on high-level goals, presenting both opportunities and risks for enterprise security. Key issues include deceptive AI behaviours and the rise of Shadow ML (machine learning) used without security oversight. Strategies for countering these risks include adversarial testing, AI Security Posture Management, and mapping AI model behaviour in real-time. Prompt injection and manipulation can also exploit Agentic AI, requiring input sanitization, multilayered authentication, and detailed auditing. Potential AI-generated false security positives and AI usage in cybercrime also need addressing.
New TgToxic Banking Trojan Variant Evolves with Anti-Analysis Upgrades
Cybersecurity researchers have found an enhanced version of the Android malware TgToxic. First documented by Trend Micro in 2023, the banking trojan steals crypto and
