Johannesburg – If you've ever felt annoyed booking trips, searching for troubleshooting information, tuning complex supply chains, or managing an elaborate environment, you're interested in what AI agents have to offer.
AI agents are poised to reconstruct the way we get things done.
They stand up for us, manage and delegate many tasks towards comprehensive outcomes, and give people time to focus on higher value tasks.
But they also pose security risks.
What do you need to know about AI agents and how to stay safe from malicious employees and cybercriminals?
Artificial helper
In the late 1980s, Francesco Cirillo had to improve his poor time management.
He saw a tomato-shaped kitchen timer and used it to split the task into smaller work intervals. He names his technique after Pomodoro, the Italian word for tomato.
Cirillo has created an elegant model for traditional productivity techniques. Split the project into smaller tasks.
Currently, artificial intelligence is evolving this concept with AI agents.
Also known as agent AI, these AIs act on behalf of humans and systems, coordinating multiple tasks.
However, instead of following a fixed instruction set, AI agents learn to adapt their behavior.
AI agents can use the system directly, such as scheduling meetings on your behalf.
It can be much more complicated, like managing supply chain items, planning elaborate trips, or being coordinated via “narrow” AIS, coordinated via scripts.
Harvard Business Review is a technology that promises to “change many aspects of human machine collaboration.”
It's no wonder that half of companies using generated AI predict that by 2027, Deloitte will adopt Agent AI.
“Agent AI is extremely exciting and could be the biggest productivity boost that everyone expects from AI,” said Gerhard Swart, chief technology officer at Performanta's cybersecurity company.
“It's more than automation. It manages complex, dynamic goals at high speed and increases the impact that machine systems have on us. But with that, new security risks arise.”
Dangerous AI
Executives trust personal assistants.
Patients trust their doctors. The inventory controller trusts the logistics manager.
That trust is very important as those people are exposed to processes and internal mechanisms of information.
“The same applies to AI agents, and soon the biggest security concerns are flagged,” Swart said.
“Agent AI requires access to a secure system to work. It has many different systems.
“This kind of access is extremely dangerous if it falls under the influence of malicious employees or cybercriminals.”
Unless AI agents are secure and managed sufficiently, the wide reach and the extent to which we rely on them poses significant risks.
Already, major vendors like Salesforce, which have an agent AI service called AgentForce, are running bug bounty programs that help you find security gaps affecting AI agents.
Protecting AI Agents
“From a risk perspective, we can think of AI agents as similar to human users who have access to sensitive systems. This comparison helps define how to keep AI agents safe,” Swart said.
“AI agents have access to critical business data, scheduling and email applications, SAAS services and more.
“There are many things you can make decisions from that perspective, such as ensuring that you understand and limit your agent's access to your system.”
Here's how to protect your AI agent:
A powerful identity management and authentication system. Policy to encourage and guide use and avoid shadow AI agents. Regular human loop assessment of AI agents output and behavior. Limit data exposure and use encryption smartly. Continuous security monitoring.
Continuous monitoring is one of the most important measures. The best managed security service providers (MSSPs) employ sophisticated surveillance and remediation systems, including AI.
Additionally, world-leading MSSPs such as Performanta take a risk-first approach to identify, monitor and protect the most important areas of their customers, reducing the associated risks and costs.
“Affiliation with MSSP is particularly wise for AI agents. MSSPS is investing in aggressive surveillance and repair tools, including AI agents that can act as fast as attackers,” Swart said.
“Active monitoring and repairs are very important and need to be very fast. These must be as fast as the machine can act. MSSP invests in this speed and accuracy.
“Partnering with top MSSPs is the best option, especially if you want to make the most of your AI agents while keeping the risks and costs low.”
