Topic starter
24/06/2026 11:58 pm
TL;DR: AI systems widen the attack surface through prompt injection, jailbreaking, deepfakes, data poisoning, and AI-generated phishing that can bypass conventional controls, according to WitnessAI. The governance problem is not just adversarial content, but the fact that AI systems can be manipulated into taking actions, handling data, or impersonating identities outside intended guardrails.
Editorial analysis by NHI Mgmt Group, based on content published by WitnessAI: “AI Cybersecurity Threats: How Artificial Intelligence Is Changing the Cyber Risk Landscape”.
Key questions
Q: What breaks when prompt injection reaches a tool-using AI agent?
A: What breaks is the assumption that the model's output is low impact.
Q: Why do AI-powered phishing and deepfakes increase credential risk?
A: They increase risk because attackers can tailor messages, voices, and context to look legitimate at scale.
Q: How should security teams decide what authority an AI system should have?
A: Start from business purpose, not technical possibility.
Practitioner guidance
- Define AI system intent boundaries Document what each AI system is allowed to summarise, retrieve, generate and trigger, then revoke any capability not required for the business task.
- Inspect external content paths for prompt injection Review web, email, document and retrieval sources that can feed prompts or context into AI systems, and block untrusted instruction-bearing inputs.
- Separate AI access from AI trust Do not let a system’s authenticated connection to data or tools imply approval to act on those resources without additional governance checks.
Bottom line: AI expands cyber risk by turning text, voice and image manipulation into identity compromise and fraudulent action.
Explore further
View Full Forum → | NHI Foundation Course → | Our Services → | Read the full analysis →
25/06/2026 8:47 am
AI has turned identity assurance into an intent problem: traditional IAM controls can verify who authenticated, but they cannot on their own prove that an AI system will stay within the purpose for which access was granted. Prompt injection and jailbreaking create a gap between permission and intent, which is why runtime governance matters as much as enrolment or authentication. Practitioners should treat AI behaviour as part of the identity boundary, not an afterthought.
A question worth separating out:
Q: How should security teams secure AI agents before they are embedded into business workflows?
A: Security teams should treat AI agents as a distinct identity and access problem, not just an application feature. The first step is to inventory agent activity, define allowed actions, and constrain access to only the data and systems each agent needs. Governance should include monitoring, policy enforcement, and review of agent decisions so unintended actions are detected early.
👉 Read our full editorial: AI cybersecurity threats are widening enterprise identity risk