The threat environment has evolved at a pace that outstrips that of corporate risk committees. Companies are dealing with unplanned AI adoption, a stretched skills base and a regulatory environment under POPIA that expects a level of data stewardship many have not yet been able to operationalise.
Trust has become central to security mandates and conversations, primarily because AI has broken the assumption of trust within the organisation. For years, this was based on recognition and familiarity – faces, writing styles, voices – but these are no longer a baseline when voice cloning, deepfakes and synthetic identities can fool trained employees. The idea that an e-mail or a call ‘looks and sounds’ like a specific person can no longer be treated as a form of verification, and this means that trust has to be re-engineered into how decisions are authorised.
Trust has become a system
AI-generated phishing has reached click-through rates of 30%-44% against the 19%-28% that traditional attempts achieve. The economics behind this change are staggering – recent testing has found that AI agents running sophisticated offensive attacks can do so for under $50. This is work that once required a human testing team that cost upwards of $100 000. It is an economics-based arms race that cannot be won by outspending the attacker, especially when their cost of experimentation is close to zero.
The answer is to change the economics by defending what matters disproportionately well and by collaboration across systems, companies and solutions. Shared threat intelligence allows smaller companies to bring their defensive economics closer to those of larger enterprises. And this aligns with the growing demand for third-party service providers, usually smaller companies, to have enterprise-level security to minimise the risk of third-party vulnerabilities.
The same change in trust has also changed the dynamic of the hack as hackers, sourcing data, login credentials and information from multiple sources and sites, now log in to companies rather than break in. The years spent building sophisticated walls around the network are being taken down by access simplified by AI and credential theft. Multi-factor authentication is now essential to every company’s security to ensure that trust is proven from the moment an individual gains access.
The gap between adoption and control
Companies are deploying models, copilots and agents faster than they are securing them, and often these tools have no clear ownership for risk. Data governance is close behind, with sensitive information fed into pipelines without a full understanding of its location and storage.
At the Armata Cybersecurity Leadership Summit 2026, the focus is on how security has become a journey built on a partnership between the CIO, CISO and the business. The event is designed to provide leaders with insight into structured responses, the risk of the AI agent, establishing ownership over AI and filling in the gaps between IT, security, legal and the teams using the tools.
The summit isn’t built around fear; it is structured to enhance and transform security decision-making, and every session is designed to move the business from consideration to planning.
The event is taking place on 29 July 2026.
"Digital trust used to rest on recognition, a familiar voice, a familiar face, a familiar writing style. AI has broken that assumption completely. Trust in this new environment isn't a feeling. It's a system.
"You can't win an economics-based arms race by trying to out-spend an attacker whose cost of experimentation is close to zero. The response has to change the economics, not just the budget line," says Caesar Tonkin, Managing Director at Armata Cyber Security.
For more information about Armata's cyber security solutions and thought leadership initiatives, reach out to the Armata team here.
Share
Editorial contacts