AI agents execute a series of tasks to achieve a goal. Because agents and the underlying model cannot reliably distinguish instructions from related content, attackers can hijack the agent's goal to pursue adversarial objectives.

Agents can misuse legitimate tools due to prompt injection, misalignment or unsafe/ambiguous delegation, leading to data exfiltration, tool-output manipulation or workflow hijacking.

Identity and privilege abuse exploits dynamic trust and delegation in agents to escalate access and bypass controls by manipulating delegation chains, role inheritance, control flows and agent context (including cached credentials).

Agentic supply chain vulnerabilities arise when agents, tools and related artefacts are provided by third parties and may be malicious, compromised or tampered with in transit (models, weights, tools, plugins).

Agentic systems often generate and execute code; attackers exploit code-generation features or embedded tool access to escalate into remote code execution (RCE), local misuse or exploitation of internal systems.

Agentic systems rely on stored and retrievable context (memory, conversation history, retrieved data); poisoning that context corrupts continuity and reasoning across tasks.

Multi-agent systems depend on continuous communication between autonomous agents coordinating via APIs, message buses and shared memory, significantly expanding the attack surface and undermining perimeter-based security.

Agentic cascading failures occur when a single fault (hallucination, malicious input, corrupted tool or poisoned memory) propagates across autonomous agents, compounding into system-wide harm and bypassing stepwise human checks.

Agents can establish strong trust with users through fluency, emotional intelligence and perceived expertise (anthropomorphism); adversaries or misaligned designs exploit this trust to influence decisions or extract sensitive information.

Rogue agents are malicious or compromised AI agents that deviate from their intended function or authorized scope, acting harmfully, deceptively or parasitically within multi-agent or human-agent ecosystems.