Cybersecurity is entering a new phase as artificial intelligence shifts from being a tool to becoming an active participant in enterprise operations, according to cybersecurity firm Palo Alto Networks.

During an online ASEAN media briefing, the company said organizations are now dealing with an “agentic” environment where AI systems can act autonomously, significantly expanding both capabilities and risks. 

Erik Papir, senior director for ASEAN technical solutions, said enterprises are moving beyond human-led workflows toward AI agents that can execute tasks across systems, applications, and networks.

“We are moving from humans to AI agents acting on behalf of humans,” Papir said.

This shift is reshaping enterprise architecture, with AI applications, enterprise agents, and agent-based endpoints and browsers becoming part of daily operations. As a result, the attack surface has expanded, requiring new approaches to security.

“For a business, this means securing the data that goes into the model and ensuring that the output is not manipulated,” Papir added.

The rise of agentic AI comes as Southeast Asia faces a shortage of cybersecurity talent, prompting companies to rely more on automation to manage threats.

“In ASEAN, we face [a] critical talent shortage. Agents are kind of our new workforce,” Papir said, warning that the same systems designed to improve efficiency can also introduce vulnerabilities.

“An agent with the power to do all this also has the power to leak.”

To address these risks, Palo Alto Networks is pushing a unified, AI-native platform designed to move cybersecurity from reactive to autonomous operations.

Central to this approach is Prisma AIRS 3.0, a platform built to secure AI systems across their lifecycle, including discovery, risk assessment, and real-time protection.

The system can identify AI agents operating across cloud and software environments, scan them for vulnerabilities, and apply automated fixes. It also simulates attacks through “red teaming” to test how systems behave under pressure.

The company also highlighted Prisma Browser, a secure enterprise browser designed to protect what it described as the modern workplace, where most work now happens online.

According to the firm, 85 percent of work occurs in the browser, while 95 percent of organizations report browser-based attacks. The platform integrates malware protection, data loss prevention, and secure access to AI tools.

In one demonstration, an AI agent was shown autonomously updating records and processing workflows within a browser environment, while security systems intercepted a hidden prompt injection attack before it could compromise data.

“It’s autonomy by default, with control where it matters,” Papir said.

Beyond AI security, Palo Alto Networks is also focusing on digital trust, particularly as certificate lifecycles shorten and identity becomes a critical security layer.

“Identity is the new perimeter,” Papir said, noting that threats increasingly stem from compromised credentials and weak access controls.

The company said it is investing in automated certificate management and identity-based security as part of a broader effort to future-proof enterprise systems.