In 1968, General Motors introduced the world’s first integrated body-colored, rubber front bumpers to the Pontiac GTO. At the time, they were unique, as it was the norm to have external chrome bumpers.
But General Motors changed the way that every other manufacturer thought about car design. At the time, the GTO’s integrated bumper and body design gave “direction and impetus for the entire industry.” Complying with safety standards was no longer something to be bolted on at the end of the process. It became part of the design process itself rather than an afterthought. Today, that kind of integration feels standard.
A similar enterprise IT moment has arrived. For decades, these were designed primarily for operational performance and scalability, with compliance treated as an issue to address later. Yet compliance encompasses regulatory, legal, recordkeeping, and supervisory obligations, as well as internal policies governing system use.
A critical juncture
Governance is the set of people, policies, and processes that ensure AI is used accountably and consistently. Retention is the practice of keeping the records and communications an organization needs, and supervision is the practice of reviewing and monitoring activity to ensure those standards are met. All are too often treated as downstream rules and safeguards governing the use and oversight of AI and enterprise communications.
However, this model will fail as AI becomes so deeply embedded in how employees communicate, generate content, and make decisions. Compliance needs to be part of the core design of every enterprise IT architecture. For example, automatically retaining AI-assisted communications and applying review requirements to higher-risk AI uses.
Otherwise, organizations put the success of those AI projects at risk. Because the ability to govern, audit, and explain AI-influenced activity increasingly determines whether organizations can put AI tools into active use at all.
When communication and decision oversight become native to architecture design, activity can be traceable, policies can be enforced in real time as work happens, and evidence is audit-ready by default.
AI changes enterprise communication
Supervision of humans by humans was already a complex challenge for organizations, but AI introduces a fundamentally different dynamic. Communications such as emails, chats, meeting summaries, reports, translations, and client-facing drafts can now be generated, summarized, translated, or influenced at machine speed.
This introduces new challenges that traditional regulatory compliance frameworks weren’t designed to answer. Organizations must produce records of human-created interactions and reconstruct the context of AI interactions, including how information was generated, what influenced it, and whether appropriate oversight was in place.
Regulatory expectations evolve
At the same time, regulators are placing greater emphasis on demonstrable governance. Organizations are increasingly expected to show that appropriate supervision, safeguards, and accountability mechanisms are in place and operating on an ongoing basis.
For example, FINRA requires financial firms to “establish and maintain” a supervisory system designed to ensure compliance. As AI systems influence customer interactions, financial guidance, and operational decisions, regulators must be able to trust the records and the integrity of the systems that produced them. Governance has to be built into the system’s very architecture.
Readiness for enterprise AI deployment
Much of the conversation around enterprise AI focuses on technical capability, including model performance, infrastructure readiness, and integration complexity. But compliance readiness is often the limiting factor.
When the legal department steps in, you must show that a new AI tool can be deployed without compliance risk, or it often can’t be deployed at all.
AI should be supervised, explained and defended in exactly the same way as any other content source. To test this, enterprise leaders can implement four foundational disciplines:
- Create a risk-tiered AI use policy: Highlight allowed use cases, data types, and whether human review is required. This means separating lower-risk uses, such as drafting internal summaries, from higher-risk uses, like customer-facing communications, or regulated activity, and assigning different review requirements to each.
- Capture evidence trails by default: Track user identity, tools used, key inputs/outputs, and data sources.
- Establish clear AI-enabled workflow accountability: Define who monitors activity, who approves higher-risk uses, and who is responsible for addressing problems, documenting the response, and preventing the issue from happening again.
- Require higher-risk pre-deployment assurance: Create post-deployment monitoring and incident playbooks for reference.
These AI systems need to be fully integrated into compliance workstreams, combining them with human-generated content to get a complete, contextualized understanding of what has happened.
This framework serves as the compliance architecture blueprint, with governance as the mechanism that enables AI deployments to scale safely.
Governance at the speed of enterprise activity
Cross-enterprise communication patterns are evolving rapidly as new collaboration platforms emerge, regulatory expectations shift, and AI capabilities continue to expand.
That speed is already creating governance blind spots. As of 2025, Gartner found that 69% of organizations saw their employees use prohibited GenAI tools. Gartner also predicts that by 2030, more than 40% of enterprises will experience security or compliance incidents linked to unauthorized AI use.
Compliance processes have to keep up. Historically, compliance infrastructure was updated periodically, aligned to software release cycles, regulatory review periods, or internal policy updates. That model assumed a relative stability that no longer exists.
What changes now is the supervision model. Enterprises need to implement oversight across all channels and AI tools. Compliance transitions from a periodic function to a continuous system, which triggers a structural evolution in how organizations manage accountability.
Compliance by design is now core enterprise infrastructure
As AI continues to reshape enterprise operations, compliance takes on a new role beyond meeting regulatory obligations. Compliance becomes a foundational infrastructure that enables organizational operations in increasingly complex environments.
This shift reflects a broader change in how organizations define operational readiness. Enterprises need to think about more than just performance and scalability when deploying new technologies—they also need to consider governability. Trust—whether from regulators, customers, or stakeholders—depends on this.
To succeed, organizations need to make governance a core part of their architecture, so that auditability and supervision operate continuously to support innovation rather than constrain it.
Soniya Bopache is the senior vice president and general manager at Arctera.