How agentic AI will pave the way for a new era of cyberattacks

As artificial intelligence becomes increasingly autonomous, Gartner forecasts that one-third of enterprise software will incorporate Agentic AI by 2028, bringing both opportunities and cybersecurity challenges.

Key predictions and adoption trends: Gartner’s latest analysis reveals significant shifts in how AI will autonomously handle business operations within the next few years.

• By 2028, 33% of enterprise software will integrate Agentic AI capabilities
• AI agents are expected to conduct 20% of digital storefront interactions
• Autonomous systems will make approximately 15% of day-to-day operational decisions

Understanding Agentic AI: This emerging technology leverages large language models (LLMs) to automate processes and make decisions independently, similar to having an AI-powered assistant that can plan and execute tasks without human intervention.

• Agentic AI combines LLM knowledge with decision-making capabilities to drive process automation
• These systems can interpret context, plan actions, and execute tasks autonomously
• The technology promises to streamline business operations and enhance efficiency across various sectors

Security implications and vulnerabilities: Gartner analysts Gary Olliffe and Jeremy D’Hoinne highlight several critical security concerns associated with Agentic AI deployment.

• Smart malware could leverage AI agents to conduct more sophisticated and automated attacks
• Prompt injection attacks may manipulate AI systems into performing unauthorized actions
• The autonomous nature of these systems amplifies potential security risks due to their ability to execute actions at scale
• Systems are vulnerable to hallucinations and misinterpretation of instructions, which could lead to operational incidents

Recommended security measures: Organizations need to implement comprehensive safeguards to protect against potential threats while leveraging Agentic AI benefits.

• Deploy multi-layered protection strategies spanning LLM, enterprise, and agent levels
• Establish strict limitations on agent capabilities and operational scope
• Implement robust input/output controls to monitor and regulate AI actions
• Combine rule-based monitoring for LLMs with policy enforcement mechanisms for agent actions

Looking ahead: While Agentic AI presents transformative potential for enterprise operations, its rapid adoption trajectory demands careful consideration of security implications and proactive implementation of protective measures to mitigate emerging risks.