We have entered the age of the human-AI hybrid, where trust becomes the primary attack surface, says Javvad Malik, lead CISO advisor at cyber security firm KnowBe4.
Malik is scheduled to deliver a keynote address at ITWeb Security Summit Cape Town 2026 on 26 May at the Century City Conference Centre.
In his presentation, Malik will show how ‘helpful’ AI can be steered into endorsing shaky assumptions, validating questionable credentials and drafting persuasive (but flawed) arguments.
He will detail how, after receiving a London parking penalty, he used AI to challenge the fine – producing documentation convincing enough to slip past modern checks and processes.
Malik will explain how, in under 48 hours, he used off-the-shelf models and agent workflows to generate a credible appeal, a coherent backstory and the kind of polished digital presence that automated decision-making increasingly treats as legitimate.
“Then we’ll zoom out from the absurd to the enterprise reality: how the same mechanics enable insider risk, process bypass and high-impact errors when organisations outsource judgment to AI summaries and automated workflows,” he added.
In this session, delegates will gain insight into how to secure this human-AI hybrid environment.
They will hear about:
- The helpful-AI trap: How agentic systems can be nudged to prioritise completion over correctness and what that means for policy and compliance.
- The summary risk: Why AI “executive summaries” can quietly introduce false confidence, flawed logic and material business decisions.
- Context as a weapon: How benign content can be reframed by AI to create convincing narratives against individuals, teams or organisations.
- Trust boundary controls: A practical framework for moving from “don’t click” to “don’t blindly follow”, with guardrails for humans and agents.
Ros Hinchcliffe, events manager at ITWeb, said the keynote represents the kind of information and insight the market needs. “We anticipate a very strong response from local delegates because we’ve all heard the narrative that AI is a double-edged sword – that it can be used both in attack and defence. Now we have real-world examples of how the technology can be manipulated to achieve specific outcomes, malicious or otherwise.
"Essentially, the more business leaders familiarise themselves with the intricacies and dynamics of AI, the more effective they will be in harnessing its advantages and being aware of its dangers.”
For more information and to register, visit ITWeb Security Summit Cape Town 2026.
Share