The Next Big AI Business: Fixing What We Failed to Govern

4 hours ago

0 6 minutes read

STORY INLINE POST

The Next Big AI Business: Fixing What We Failed to Govern Jalisco Renews Tech Hub Act, Expands Industry Modernization South Korea Proposes US$597 Billion Budget to Accelerate AI How to Turn Your Wi-Fi Into a Strategic Revenue Channel Why the Workplace Is Mexico's Front Line for Preventive Health Sony, Warner Sue Anthropic Over Claude AI Music Training Breadth Over Depth: Why Generalists Win in the Age of AI AI Agents Bring Governance, Efficiency to Financial Institutions Idioma Leer en Español DIA assistant 1.0x ✕

Artificial intelligence is beginning to feel oversaturated. Almost every product is now AI-powered, AI-enabled, AI-native, or agentic. Companies that three years ago sold software, analytics, or automation increasingly describe themselves as AI companies. The language appears everywhere, from investor presentations and strategic plans to corporate announcements and explanations for workforce reductions.

Not all of this is empty positioning. AI is already reshaping industries, workflows, and business models. But genuine transformation is unfolding alongside something else. AI washing has become increasingly difficult to ignore, with organizations exaggerating the role, maturity, or impact of artificial intelligence in their operations.

The market is beginning to distinguish more clearly between substance and narrative. A study published in Finance Research Letters in April 2026, based on US companies between 2018 and 2023, found that unsupported AI claims can initially attract investor attention but later trigger negative market reactions and weaker performance. Gartner added another warning on July 22, 2026, noting that less than 1% of layoffs in 2025 were directly tied to AI productivity gains, despite the prominence of AI in many corporate explanations for restructuring.

Beyond the AI Saturation Point

None of this means AI is becoming less relevant. The opposite may be true. Stanford’s "AI Index Report 2026" found that 88% of surveyed organizations already use artificial intelligence, while 70% use generative AI in at least one business function. AI is no longer confined to innovation labs or isolated pilots. It is becoming embedded in daily operations.

The more important problem is that adoption appears to be moving faster than the ability of organizations to understand and govern what they are deploying. On July 8, 2026, Gartner reported that only 21% of C-suite executives could be considered genuinely AI-savvy, while 91% admitted to overstating how much they know about artificial intelligence. On June 15, Gartner reported that 71% of CEOs believe their current IT operating models are not fit for the AI era.

That contradiction should concern us more than the hype itself. Companies are deploying increasingly powerful technologies inside organizations whose own leaders acknowledge that they do not fully understand them and whose operating structures were designed for a different technological era.

On Aug. 26, 2026, Bill Gates framed the problem in broader terms. Writing about what he described as a turbulent AI era, he argued that leaders, experts, and communities are not preparing adequately for the transition. His assessment was uncomfortable in its simplicity. “There is no plan to ease the entry into the AI era.”

Governance Starts Before the Machine

Recently, ISDI invited me to teach a class on AI governance, risk, and ethical use. My intention was not to overwhelm the session with the European AI Act, Brussels, Hiroshima, the G7, or OECD principles. Those frameworks matter, but we sometimes treat them almost as the altars of governance and forget that governing technology begins much earlier.

Governance starts before anyone touches a machine, trains a model, or deploys an agent. It begins with design and with identifying the object of governance itself. An organization first needs to understand what it is trying to govern, which outcome it wants to protect, what behavior it is willing to accept, and which consequences it wants to avoid.

The simplest way I found to explain this was through individual behavior. We follow small rules every day without describing them as governance. We sleep, eat, rest, drink water, and respect certain limits because we understand what happens when we do not. We can ignore those rules for a while and apparently get away with it. Eventually, the body presents the bill.

Organizations can do the same. A company can connect a model without documenting it, send information to a provider without fully understanding how it will be processed, automate a workflow without assigning clear responsibility, tolerate unauthorized tools, or deploy agents with excessive permissions. For months, perhaps nothing visibly goes wrong. Until something does.

Stanford is already documenting signs of that gap. Reported AI-related incidents increased from 233 in 2024 to 362 in 2025. At the same time, the AI Index warns that responsible AI practices are not advancing at the same pace as capabilities and deployment.

The Remediation Market Is Already Forming

This is where one of the next major business opportunities in artificial intelligence begins to emerge.

The first wave of value went to those building models, chips, infrastructure, and platforms. Another enormous opportunity emerged for companies helping organizations adopt the technology. Governments, regulators, and international institutions then began defining the boundaries within which AI would operate.

The next opportunity may belong to those who walk into organizations after the adoption rush and determine what was done incorrectly.

This market is no longer theoretical. On Aug. 12, 2026, Deloitte announced an expansion of its AI Controls and Assurance capabilities as companies prepared to scale autonomous agents. The firm cited a striking gap. Seventy-four percent of companies plan to deploy agentic AI within two years, but only 21% report having a mature governance model for autonomous agents. PwC is already offering dedicated Assurance for AI services, while KPMG markets AI inventories, risk assessments, governance frameworks, controls, and assurance.

The race is therefore not only about who can deploy AI faster. A parallel market is already forming around who can make those deployments governable, auditable, explainable, and defensible once they become embedded in the business.

Many companies will not simply need another policy document or an ethics committee. They will need to reconstruct inventories of AI systems, identify data flows, review vendors, restore traceability, restrict permissions, examine automated decisions, establish accountability, test controls, and remediate regulatory, operational, or reputational exposure. In some cases, this work will have to be done while the systems remain in production because the organization has already become dependent on them.

That is fundamentally harder than governing correctly from the beginning. It will also be considerably more expensive.

When Consequences Become a Business

Cybersecurity offers a useful precedent. Companies first connected systems, employees, customers, and data. Only afterward did we build an enormous industry to protect everything that had been connected. AI may be following a similar pattern at much greater speed. We are first automating, integrating, and delegating. Later, many organizations will have to recover control, explainability, and accountability.

This is why governance could become one of AI’s next major industries, but not simply because regulation is expanding or because companies will suddenly develop a passion for ethics. It will become a business because consequences have a price.

Some of the first fortunes of the AI era will belong to those who built the technology. Others will belong to those who successfully scaled it. But another generation of value is already beginning to form around the people and firms capable of entering later, understanding what happened, and helping organizations repair what they failed to govern correctly from the beginning.

The next great question in artificial intelligence may no longer be what we can build. It may be who can help us regain control of what we already built.