Your AI use policy is solving the wrong problem

When a company with tens of thousands of software engineers found that uptake of a new AI-powered tool was lagging well below 50%, they wanted to know why. It turned out that the problem wasn’t the technology itself. What was holding the company back was a mindset that saw AI use as akin to cheating. Those who used the tool were perceived as less skilled than their colleagues, even when their work output was identical. Not surprisingly, most of the engineers chose not to risk their reputations and carried on working in the traditional way.

These kinds of self-defeating attitudes aren’t limited to one company—they are endemic across the business world. Organizations are being held back because they are importing negative ideas about AI from contexts where they make sense into corporate settings where they don’t. The result is a toxic combination of stigma, unhelpful policies, and a fundamental misunderstanding of what actually matters in business. The path forward involves setting aside these confusions and embracing a simpler principle: Artificial intelligence should be treated like any other powerful business tool.

This article shares what I have learned over the past six months while revising the AI use policies for my own companies, drawing on the research and insights of my internal working group (Paul Scade, Pranay Sanklecha, and Rian Hoque).

Confusing Contexts

In educational contexts, it is entirely appropriate to be suspicious about generative AI. School and college assessments exist for a specific purpose: to demonstrate that students have acquired the skills and the knowledge they are studying. Feeding a prompt into ChatGPT and then handing in the essay it generates undermines the reason for writing the essay in the first place.

When it comes to artistic outputs, like works of fiction or paintings, there are legitimate philosophical debates about whether AI-generated work can ever possess creative authenticity and artistic value. And there are tough questions about where the line might lie when it comes to using AI tools for assistance.

But issues like these are almost entirely irrelevant to business operations. In business, success is measured by results and results alone. Does your marketing copy persuade customers to buy? Yes or no? Does your report clarify complex issues for stakeholders? Does your presentation convince the board to approve your proposal? The only metrics that matter in these cases are accuracy, coherence, and effectiveness—not the content’s origin story.

When we import the principles that govern legitimate AI use in other areas into our discussion of its use in business, we undermine our ability to take full advantage of this powerful technology.  

The Disclosure Distraction

Public discussions about AI often focus on the dangers that follow from allowing generative AI outputs into public spaces. From the dead internet theory to arguments about whether it should be a legal requirement to label AI outputs on social media, policymakers and commentators are rightly concerned about malicious AI use infiltrating and undermining the public discourse.

Concerns like these have made rules about disclosure of AI use central to many corporate AI use policies. But there’s a problem here. While these discussions and concerns are perfectly legitimate when it comes to AI agents shaping debates around social and political issues, importing these suspicions into business contexts can be damaging.

Studies consistently show that disclosed AI use triggers negative bias within companies, even when that use is explicitly encouraged and when the output quality is identical to human-created content. The study mentioned at the start of this article found that internal reviewers assessed the same work output to be less competent when they were told that AI had been used in its production than when they were told it had not been, even when the AI tools in question were known to increase productivity and when their use was encouraged by the employer. Similarly, a meta-analysis of 13 experiments published this year identified a consistent loss of trust in those who disclose their AI use. Even respondents who felt positively about AI use themselves tended to feel higher distrust toward colleagues who used it.

This kind of irrational prejudice creates a chilling effect on the innovative use of AI within businesses. Disclosure mandates for the use of AI tools reflect organizational immaturity and fear-based policymaking. They treat AI as a kind of contagion and create stigma around a tool that should be as uncontroversial as using spell-check or design templates—or having the communications team prepare a statement for the CEO to sign off on.

Companies that focus on disclosure are missing the forest for the trees. They have become so worried about the process that they’re ignoring what actually matters: the quality of the output.

The Ownership Imperative

The solution to both context confusion and the distracting push for disclosure is simple: Treat AI like a perfectly normal—albeit powerful—technological tool, and insist that the humans who use it take full ownership of whatever they produce.

This shift in mindset cuts through the confused thinking that plagues current AI policies. When you stop treating AI as something exotic that requires special labels and start treating it as you would any other business tool, the path forward becomes clear. You wouldn’t disclose that you used Excel to create a budget or used PowerPoint to design a presentation. What matters isn’t the tool—it is whether you stand behind the work.

But here’s the crucial part: Treating artificial intelligence as normal technology doesn’t mean you can play fast and loose with it. Quite the opposite. Once we put aside concepts that are irrelevant in a business context, like creative authenticity and “cheating,” we are left with something more fundamental: accountability. When AI is just another tool in your tool kit, you own the output completely, whether you like it or not.

Every mistake, every inadequacy, every breach of the rules belongs to the human who sends the content out into the world. If the AI plagiarizes and you use that text, you’ve plagiarized. If the AI gets facts wrong and you share them, they’re your factual errors. If the AI produces generic, weak, unconvincing language and you choose to use it, you’ve communicated poorly. No client, regulator, or stakeholder will accept “the AI did it” as an excuse.

This reality demands rigorous verification, editing, and fact-checking as nonnegotiable components of the AI-use workflow. A large consulting company recently learned this lesson when it submitted an error-ridden AI-generated report to the Australian government. The mistakes slipped through because humans in the chain of responsibility treated AI output as finished work rather than as raw material requiring human oversight and ownership. The firm couldn’t shift blame to the tool—they owned the embarrassment, the reputational damage, and the client relationship fallout entirely.

Taking ownership isn’t just about accepting responsibility for errors. It is also about recognizing that once you have reviewed, edited, and approved AI-assisted work, it ceases to be “AI output” and becomes your human output, produced with AI assistance. This is the mature approach that moves us past disclosure theater and toward genuine accountability.

Making the Shift: Owning AI Use

Here are four steps your business can take to move from confusion about contexts to the clarity of an ownership mindset.

1. Replace disclosure requirements with ownership confirmation. Stop asking “Did you use AI?” and start requiring clear accountability statements: “I take full responsibility for this content and verify its accuracy.” Every piece of work should have a human who explicitly stands behind it, regardless of how it was created.

2. Establish output-focused quality standards. Define success criteria that ignore creation method entirely: Is it accurate? Is it effective? Does it achieve its business objective? Create verification workflows and fact-checking protocols that apply equally to all content. When something fails these standards, the conversation should be about improving the output, not about which tools were used.

3. Normalize AI use through success stories, not policies. Share internal case studies of teams using AI to deliver exceptional results. Celebrate the business outcomes—faster delivery, higher quality, breakthrough insights—without dwelling on the methodology. Make AI proficiency a valued skill on par with Excel expertise or presentation design, not something requiring special permission or disclosure.

4. Train for ownership, not just usage. Develop training that goes beyond prompting techniques to focus on verification, fact-checking, and quality assessment. Teach employees to treat AI output as raw material that requires their expertise to shape and validate, not as finished work. Include modules on identifying AI hallucinations, verifying claims, and maintaining brand voice.

The companies that will thrive in the next year won’t be those that unconsciously disincentivize the use of AI through the stigma of disclosure policies. They will be those that see AI for what it is: a powerful tool for achieving business results. While your competitors tie themselves in knots over process documentation and disclosure theater, you can leapfrog past them with a simple principle: Own your output, regardless of how you created it. The question that will separate winners from losers isn’t “Did you use AI?” but “Is this excellent?” If you’re still asking the first question, you are already falling behind.