AI Is Rewriting Security’s Job: Your Org Chart Hasn’t Caught Up
AI is forcing CISOs to reconsider what their organizations are designed to do. Security teams must protect systems, data, and assets while also demonstrating that AI-driven systems operate within approved boundaries. Fulfilling those responsibilities requires CISOs to rethink their teams’ roles, skills, and relationships with the rest of the business. Our latest research, Introducing The Trust And Assurance Organization (Forrester clients only), examines how the cybersecurity function needs to evolve and provides a starting point for CISOs planning that transition.
AI Is Breaking Security’s Old Operating Model
Agentic AI is becoming part of enterprise operations, even as privacy, security, and governance concerns complicate adoption. Meeting those concerns takes more than just new tools or policies; it requires security leaders to challenge assumptions about how security work is organized, assigned, and evaluated, as well as address changes head on. This includes the fact that:
- The business is deploying agents faster than it can govern them. In Forrester’s State Of AI Survey, 2026, 37% of C-level AI decision-makers at enterprises said that six or more of their AI and agentic projects failed before reaching production. Among the greatest barriers to adoption, they cited data privacy and security concerns, governance and risk, and trust in agentic systems. Yet boards and executives continue pursuing agents because the economic incentives favor automation, productivity, and scale. This rate of adoption, and the need for trust, expands what security has to do: CISOs must produce defensible evidence that agents, data, models, and automated decisions stay within policy, even when the business units deploying them sit outside security’s control.
- Security’s own work is also being automated at an unprecedented pace. As agents take on detection, triage, and response tasks, security teams must determine what those systems can do, when people need to intervene, and how to validate the results. Richard Watson, global cybersecurity consulting leader at EY, told us that, for CISO organizations, “the onus is on them to assure that the work is automated rather than doing it themselves.” This shift changes the skills, responsibilities, and structure the security function needs.
Evolve Your Remit To Produce Assurance And Earn Trust
Business leaders still own business decisions and accept the remaining risk, while CISOs assume the responsibility for demonstrating that security controls work. AI broadens the scope of that responsibility to include automated systems used across the business and within security itself. Security teams must establish requirements, enforce them through technical controls, validate the results, and produce evidence that stakeholders can examine. Doing that work requires the authority, skills, and organizational structure to support it.
Enter the trust and assurance organization (TAO), which Forrester defines as:
The CISO-led evolution of the cybersecurity function for an era in which digital and AI-driven systems make, recommend, and automate business decisions to protect systems, data, and assets while demonstrating that they operate securely, responsibly, and as intended across the value chain. It uses verifiable controls, governance, evidence, and transparent reporting to attest to the integrity, security, privacy, and responsible use of data and AI and enables stakeholders to determine whether the organization is meeting its obligations.
The difference between trust and assurance matters. Assurance is the mechanism, which is what the security organization produces: technically enforced controls, continuous validation, and evidence that stakeholders can examine. Trust is the outcome, which is what stakeholders decide after they examine the evidence, and it’s why security can earn trust but never certify it.
Test Your Security Organization For Authority And Capability Gaps
To see where you and your current organization stand in terms of TAO capabilities, trace one production AI use case and ask three questions: Who owns the business outcome and accepts the residual risk? Who has the authority to change the agent’s permissions and operating boundaries? What current evidence shows that those boundaries hold? Apply the same logic to one of your automated security workflows. The answers will expose one or both of two gaps:
- An authority gap. This is where security is expected to explain an outcome but has no say over the decisions and controls that produced it. Because that gap is a matter of decision rights, to close it, you need to have a discussion with your executive peers about risk and control ownership, as well as reporting lines.
- A capability gap. In this gap, security has the right authority but can’t produce the evidence because its functions still depend on periodic reviews and manual handoffs. To close this gap, you must urgently change how the security organization itself is structured, staffed, and measured.
Build The TAO Before Someone Else Builds It For You
Enterprises are already redistributing responsibility for AI governance, identity, data protection, and customer trust, which means roles left undefined get assigned through momentum rather than design. You can’t afford to wait for your responsibility to be handed out to you. Use our research, which provides you with a reference organizational chart, explains how eight security functions change, sets out the skills and career paths your teams will need, and replaces activity counts with metrics that show whether controls work.
Forrester clients can read Introducing The Trust And Assurance Organization and schedule a guidance session or inquiry with Madelein or me to discuss where their organization needs to change and what to tackle first.