AI Risk No Longer Stops at Your Walls
As AI becomes embedded in business, decisions made inside a handful of technology companies are shaping the opportunities and risks of organizations everywhere.
By Paula Fontana · November 10, 2026
For decades, organizations have built their approaches to governance and risk around decisions they can influence. Their people, systems, operations and the third parties they choose to work with.
AI is changing the nature of that relationship.
A relatively small number of companies are developing the foundation models that an expanding share of businesses will depend on. Decisions about how those models are trained, what they are permitted to do, how they are priced and who can access them have consequences that extend well beyond the companies making them.
In the first episode of Uncharted, Deniz Tudor raised a question that deserves more attention in executive teams and boardrooms: How much of our future now depends on the judgment of people inside companies we have little ability to influence?
When innovation concentrates decision-making power
Every major technology shift creates new dependencies. Cloud computing changed how organizations acquire and operate infrastructure. Digital platforms reshaped access to customers and markets. AI is extending this pattern into the capabilities organizations use to make decisions, develop products and perform work.
The concentration matters because foundation model providers influence far more than technical performance. Their choices about safety, acceptable use, model behavior and commercial access can determine what other companies are able to build and how reliably they can operate.
An organization might invest heavily in an AI-enabled product, only to find that a model update changes its performance, a policy revision restricts a use case or a pricing decision alters the economics of the business.
These are strategic dependencies because they can affect the organization’s ability to create and capture value.
The more deeply AI is integrated into products, operations and decision-making, the more consequential those external choices become.
The people behind the models matter
One of the more interesting points Deniz raised was the role of conduct risk.
In financial services, conduct risk examines how the behavior and decisions of people within an institution can create harm for customers, markets and the institution itself.
With foundation AI, the implications of organizational conduct can extend across entire industries.
The incentives, judgment and internal culture of the teams building these models influence how technology develops and what safeguards are prioritized. Customers inherit some of the consequences, often without meaningful visibility into how those decisions were reached.
We also discussed employees leaving major AI companies and speaking publicly about concerns over safety, governance and the direction of development.
These voices raise questions about how organizations handle internal dissent, how concerns reach decision-makers and what happens when commercial incentives and technical judgments come into tension.
For companies building on AI, these are signals worth paying attention to. The governance of a technology provider increasingly becomes part of the environment in which its customers operate.
Innovation and security now share the same dependencies
This is where the relationship between innovation and security becomes particularly important.
The same AI provider that enables a company to develop new products, accelerate operations or enter a new market can also introduce dependencies that influence its strategic freedom.
Product teams may evaluate a model based on capability, speed and cost. Security teams may examine data exposure, access controls and operational vulnerabilities. Executive teams may focus on investment returns and competitive positioning.
Each perspective captures part of the decision.
Understanding the full dependency requires connecting them. How easily could the organization change providers? What happens to its commercial proposition if model capabilities or access change? Which decisions can it control, and which has it effectively delegated to someone else?
These questions belong in the same conversation as the opportunities AI creates, because they influence how much of that opportunity an organization can sustain.
Governance needs to follow the decisions that matter
Boards and executive teams cannot govern the internal culture of every technology company they depend on. They can, however, make more deliberate decisions about where they place their dependencies and how much influence they retain.
That begins with understanding which capabilities are strategically important, where provider concentration exists and what evidence supports confidence in a provider’s governance.
It also means examining commercial flexibility, substitution options, contractual protections and the consequences of changes outside the organization’s control.
The objective is to make dependencies visible enough that leaders can evaluate them alongside the value they expect to create.
As AI becomes part of the infrastructure of business, some of the most consequential decisions affecting an organization will be made by people who do not work there.
That reality changes what responsible innovation demands of leadership. The ability to move quickly and build new capabilities increasingly depends on understanding who else has the power to shape their future.
This article extends a discussion from Episode 1 of Uncharted: Doomscrolling Is Not a Strategy, with Paula Fontana and Deniz Tudor.