Securing the Future: How Cyber Vets Are Building Trust in Agentic AI

WASHINGTON – May 18, 2026 – In a move that signals a critical shift towards proactive security in artificial intelligence, cyber incident response firm MOXFIVE has joined the Agentic AI Foundation (AAIF). The partnership aims to embed real-world, frontline cybersecurity expertise into the foundational architecture of the next wave of AI, ensuring these powerful systems are built for resilience from the ground up.

MOXFIVE, a firm known for its work in the trenches of major cyberattacks, will contribute its “breach-informed perspective” to the AAIF, a neutral, open-source initiative hosted by the Linux Foundation. The foundation is rapidly becoming the central hub for developing the open, interoperable standards that will underpin agentic AI—a technology poised to move AI from a passive tool to an autonomous actor in the digital world.

This collaboration addresses a growing concern among technologists and enterprise leaders: as AI agents gain the ability to set goals, make decisions, and execute tasks with minimal human oversight, they create a new and formidable attack surface. By integrating lessons learned from thousands of real-world security breaches, this initiative represents a concerted effort to build defenses before vulnerabilities become systemic.

A New Frontier of Autonomous Risk

Agentic AI marks a significant evolution from the generative AI models that have captured public attention. Where models like ChatGPT primarily generate content in response to prompts, agentic systems use these outputs as a 'brain' to perform actions, interact with digital environments, and execute complex workflows. These agents can manage supply chains, autonomously develop software, or handle intricate customer service tasks, often by collaborating with other agents.

This leap in capability, however, introduces a new class of security threats. Experts warn of risks unique to these autonomous systems, including:

• Prompt Injection: Attackers can embed hidden instructions into the data an agent processes, tricking it into revealing sensitive information or invoking tools for malicious purposes.
• Memory Poisoning: Because agents maintain a memory to learn and adapt, an attacker could corrupt this long-term state, leading to persistent, compromised behavior that is difficult to detect.
• Tool Misuse: Agents are given access to various software tools and APIs to perform their tasks. A compromised or manipulated agent could abuse these permissions, creating a cascade of unauthorized actions across an organization's digital infrastructure.

“The conversation around agentic AI has fundamentally shifted,” said Mazin Gilbert, Executive Director of the Agentic AI Foundation, in a recent statement. “Organizations building production systems are choosing to invest in open standards because they understand fragmented, proprietary approaches don’t scale. There’s growing consensus that the future of agentic AI depends on open, interoperable protocols that everyone can build on and trust.”

From Crisis Response to Resilient Code

This is where MOXFIVE’s unique expertise becomes invaluable. The firm specializes in helping organizations recover from catastrophic cyber incidents, giving them direct visibility into how threat actors operate, where defenses fail, and what is required to build true operational resilience. Instead of theorizing about potential AI vulnerabilities, MOXFIVE brings a library of knowledge from actual security failures.

“Building trust in agentic AI requires understanding how these systems perform under real-world attack conditions,” explained Ben Harel, CTO of MOXFIVE. “At MOXFIVE, we work on the front lines of cyberattacks, giving us direct visibility into how threat actors are beginning to target weaknesses in agentic AI systems and leverage these technologies for malicious purposes. We joined the AAIF to contribute that perspective and help ensure the open standards being developed today reflect the ways these systems are actually tested and attacked in the wild.”

This “security by design” philosophy is central to the AAIF’s mission. Since its launch in late 2025, the foundation has attracted over 200 member organizations, including industry giants like Google, Microsoft, AWS, OpenAI, and Anthropic. By creating a neutral space for these competitors to collaborate on foundational projects—such as the Model Context Protocol (MCP) for connecting AI models to tools and data—the AAIF aims to prevent the kind of fragmented, insecure ecosystem that has plagued other emerging technologies.

MOXFIVE’s contribution will focus on translating its post-breach knowledge into practical security controls, architectural best practices, and resilience patterns for the AAIF's open standards. The goal is to ensure that agentic systems are not only secure against known attack vectors but are also designed to withstand and recover from novel threats, with clear audit trails and fail-safes built in.

Building the Foundation for Enterprise Trust

The implications of this proactive security effort extend far beyond the technical realm. For enterprises, the decision to deploy autonomous AI agents into critical business processes hinges on one factor: trust. A single, high-profile security failure involving an agentic system could set back adoption by years, regardless of the technology's potential benefits.

By helping to bake security and resilience into the open standards that will govern agentic AI, MOXFIVE and its fellow AAIF members are laying the groundwork for widespread, responsible adoption. Secure, interoperable standards foster confidence among business leaders and IT decision-makers, de-risk investment, and prevent the vendor lock-in that stifles innovation.

Furthermore, this work is critical for addressing the inevitable governance, risk, and compliance (GRC) challenges that will accompany agentic AI. Demonstrating that these systems are built on a secure and transparent foundation, with input from frontline cybersecurity experts, will be essential for navigating regulatory scrutiny and earning public trust.

As organizations look to harness agentic AI as a defensive advantage in their own security operations, the integrity of the underlying platforms is paramount. The collaboration between a cyber crisis firm like MOXFIVE and a standards body like the AAIF represents a crucial step in a broader industry movement to secure the future of AI before it fully arrives, ensuring this powerful technology evolves into a reliable partner rather than an unpredictable new liability.