The saying goes that every action must have an equal and opposite reaction. What about no action at all?
President Trump’s new “AI Action Plan” has a message for American businesses: Go faster. Beat the competition. Win the AI race. The sweeping plan contains many specific proposals around growth and “winning the AI race” against China. However, when it comes to oversight, it essentially tells federal regulators to cool it and expressly forbids state regulators from stepping into the fray at all.
The White House is engineering a plan that emphasizes speed over sight. That has big implications for businesses and creates both an urgent risk and an unexpected opportunity for enterprise leaders.
Companies are Already Flying Blind
The numbers tell the story. Even before the federal government told everyone to plow ahead, many businesses didn’t fully understand how their workers were using AI.
Gallup research shows 40% of workers now use AI tools, while only 22% of companies have actual AI strategies. But here’s what those numbers don’t capture: Your employees aren’t trying to break rules or create security incidents. They’re trying to do their jobs better, faster, and more effectively.
As we worked together, the Chief Information Officer of one Fortune 500 company, who thought they had seven approved AI tools, discovered they actually had 34. It wasn’t unlike discovering you have 34 departments when you thought you had seven — except these “departments” were AI tools processing company data outside your management structure.
The companies I work with every day tell me similar stories: They discover pockets of their business using AI that have slipped through every filter. AI experimentation is done in so many different ways, on so many different platforms, for so many different purposes, that it can be hard to track.
The Management Blind Spot
Think about how you manage other critical business functions. You have dashboards for sales performance, operational metrics, financial health, and employee productivity. You can share exactly how many vendors you’re working with, what your cloud infrastructure costs, and who has access to sensitive customer data.
But ask most executives which AI tools their teams are using, what data those tools are processing, or what productivity gains they’re delivering, and you get blank stares.
Companies are already flying blind, in other words, and that creates a visibility crisis with three immediate management challenges:
Shadow AI proliferation: Teams are using AI tools daily, often with sensitive information that puts an organization at risk. It’ll be difficult to secure what you can’t see, and most AI usage happens through encrypted web traffic that bypasses traditional network monitoring solutions, like having vendor relationships with no contracts, oversight, or audit trail.
Competitive disadvantage accelerating: While American companies debate frameworks, competitors are building systematic AI capabilities. Europe’s AI Act is creating advantages for companies that build transparency and accountability from day one. China’s coordinated approach, which is exemplified by DeepSeek building ChatGPT-equivalent AI for $5.6 million while OpenAI spent $100 million, enables them to develop frontier AI at a fraction of the cost.
Resource misallocation: Companies are evaluating million-dollar enterprise AI platforms while most valuable use cases happen through $20/month consumer subscriptions, like a standard GPT Plus subscription, that employees discovered organically. That’s essentially spending massive budgets on enterprise software while critical work happens through unmanaged personal accounts.
As a company, you’d never let someone represent you to customers without trusting their work product. AI shouldn’t be any different.
Building an AI Management Approach
So what’s an ambitious company to do? Build a systematic AI management strategy in three steps, the same way you’d approach managing any other critical business function.
First, start with visibility, not policies: Before writing AI governance rules, understand what’s actually happening. Which tools are your teams using? What value are they creating? What risks are they generating? Remember that you can’t govern what you can’t see. And all you really need to do, to start, is see what’s going on.
Second, build outcome-based accountability: This means going deeper than CEO mandates to “use more AI” or counting AI deployments as progress. Measure productivity gains, track prompt-level usage patterns and risk exposure. Determine what you want AI to do—and measure it. One executive focused on AI implementation told me his organization has focused on “golden use cases” for AI to ensure that its most forceful efforts are in places that the organization has strategically identified AI as an accelerant. In turn, this company has chosen to avoid heavily pushing AI use cases in other areas for now.
The third step is engineering transparency systems in your organization: Remember that AI isn’t just a tool. It’s a teammate embedded in your day-to-day work. That means requiring new forms of transparency in human and AI collaborations. Companies that do this will go a step further than avoiding regulatory backlash in the long run. They’ll build competitive advantages.
The Regulatory Boomerang is Coming
The other important thing to remember: Today’s hands-off approach won’t last.
Whether it’s political winds or, more predictably, an AI incident that creates a crisis — such as a data breach, discrimination, or financial manipulation —inaction will, in fact, have an equal and opposite reaction. Stricter regulations will arrive, and companies that were forward-looking and put robust internal management and observation systems in place will do much better than those that didn’t.
The Choice Ahead
Washington has decided to sacrifice oversight for speed. Enterprise leaders don’t have to (and shouldn’t) make that bet. Organizations can build AI systems that are both powerful and trustworthy and compete on reliability, not just capability. There’s an opportunity to turn responsible AI management from a compliance cost into a competitive advantage.
Enterprise employees are already running the largest operational experiment in most organizations’ history. They’re testing tools, developing workflows, and discovering what actually drives business value. The companies that can see and manage what they’ve discovered will dominate those flying blind.
Washington eliminated AI oversight to help companies move faster. But the equal and opposite reaction to “move fast and break things” isn’t “move slow and fix things.” It’s “move fast and see things.”
The companies that crack that code first will own the advantages that actually matter for this new AI era.
