Blog Security The Edge of Agency: Defending Against the Risks of Agentic AI Maxim Zavodchik is an experienced security research leader with a proven track record in establishing, growing, and defining strategic vision for Threat Research and Data Science teams in Web Application Security and API Protection. When he’s not protecting life online, you can find him being a super dad and/or watching Studio Ghibli movies. This shift brings powerful new capabilities, but also a new class of security risks. Contents \r\n \r\n Executive summary \r\n Introduction: From automation to autonomy \r\n Why agentic AI is different \r\n Autonomy with consequences \r\n Immature security ecosystem \r\n Excessive agency and hijacking risks \r\n Lateral propagation and cascading hallucinations \r\n \r\n \r\n Autonomous access, autonomous threats \r\n Vibe scraping \r\n \r\n \r\n Resilience by design: Securing agentic application channels \r\n Akamai Firewall for AI: Specialized threat protection \r\n Advanced API security:…