Artificial intelligence is seemingly all that leaders can talk about right now. It’s discussed in all-hands meetings, included in individual and company goals, and employees are measured by how many tokens they’re using or hours spent on pilot projects. No doubt, AI is the technology du jour, and is changing how business is approached.
That said, leaders should keep their eyes open about the risks, as many are not looking at the whole picture and assessing the damage those risks can cause. We asked our Fast Company Impact Council members to share one AI risk that they think leaders underestimate. There was a surprising amount of diversity in those risks. Here is what 22 of our members said.
Much of what I’m reading and hearing focuses on accuracy, but I think the bigger risk is getting the right content in the right context. Everyone is rushing to automate workflows without measuring if the output actually resonates with the intended audience. AI may make organizations more efficient, but if it doesn’t understand human motivation and behavior mindsets, it can also make them more disconnected from the people they’re trying to reach. — Ran Mullins, Psympl
One AI risk leaders underestimate is deploying AI that produces inaccurate or inconsistent outputs at scale. The pressure to move fast can sometimes lead teams to skip the rigor needed to ensure AI performs reliably across diverse users. Inconsistent AI erodes end-user confidence in your product, so getting the fundamentals right is critical. — Alice Chang, Perfect Corp.
Many leaders focus on whether AI is accurate or fast, and tend to overlook a more important question: Are people still thinking critically? As AI becomes embedded in daily work, the greater risk is diminished judgment. Leaders must develop discernment, contextual awareness, and the confidence to challenge the machine when necessary. — Ajay Tejasvi, TLEX
The core risk leaders underestimate is over-reliance on autonomous AI in high-impact areas like health and nutrition, which can end up eroding customer trust. Deploying AI without transparency and accountability risks losing customer loyalty. The power of humans in the loop (e.g., expert review, escalation, and oversight) is the key to demonstrating credibility, building loyalty, and mitigating this black-box risk. — Tara Zedayko, Ollie Pets
Data quality is one of the most underestimated risks in AI. In a complex operation, incomplete or inconsistent data can quickly show up in recommendations that look credible on the surface but lead teams in the wrong direction. Strong AI starts with a trusted data foundation, clear accountability for data quality, and input from the people closest to the work, so the output supports better decisions and helps teams execute with confidence. — Dennis Anderson, ArcBest
The biggest risk leaders underestimate is assuming AI will fix a broken growth model. It won’t. AI won’t replace marketing; it will expose bad marketing. If your growth model only works because of long timelines, large teams, or inefficient processes, that fragility was already there. AI simply makes it visible. The good news is that for leaders willing to look at that honestly and rebuild, it’s a remarkable moment to be in the golden age of marketing. — Vineet Mehra, Chime
The race towards AI automation driven by an eagerness to reduce headcount is bound to backfire. Developing and extracting best practices while using AI has to come ahead of rolling out automation. Otherwise, the only thing agents will achieve is to automate and scale mediocrity. — Pierre-Loic Assayag, Traackr
The underestimated risk is the cost of AI. The rubber is meeting the road now, and we’re starting to see the AI companies transfer more of the cost to consumers. AI has always been expensive, and the companies providing it have been running at a steep loss. Now that we’re all hooked, and the energy and mineral resources are more constrained, the cost to consumers is increasing steadily. I believe that will continue. What’s important now is making sure teams aren’t using AI lazily or carelessly, and that requires skill and training. — Lindsey Witmer Collins, WLCM App Studio and Scribbly Books
The risk is using AI to avoid being human. AI handles execution brilliantly: the drafting, synthesis, pattern recognition, optimization. But it can’t read the actual human energy in a room, put its name on a direction without full certainty, or decide to genuinely care about something and mean it. Leaders who use AI to feel productive while avoiding the work that’s actually theirs—the judgment calls, the honest conversations, the risky, creative bets that could fail—are optimizing for the wrong thing. The tool is only as good as the human behind it. — Muneer Panjwani, Engage for Good | The Halo Awards
AI is a powerful tool. Used intentionally, it can deliver significant business value. As companies embrace AI, they must also understand the risks, such as unsecured employee use. Employees may unintentionally share sensitive information, including customer data, financial records, or trade secrets, or expose intellectual property. This can lead to security vulnerabilities, reputational damage, legal exposure, and operational risks. Organizations that benefit most from AI establish clear governance, provide employee education, and implement secure AI solutions. — David Klanecky, Cirba Solutions
I’ll give you two. First, AI is a reversion to the mean and will drag down your best talent. Be strategic in its implementation to shore up weaknesses and fill gaps, don’t replace actual talent. Second, be mindful of your incentives. If you inadvertently reward speed and volume of work over a high standard of quality, then you risk a culture of AI slop. — Bo Zhao, Baby Gear Group
We are making consequential decisions about how millions of people will experience AI, and we are making them fast, under pressure, with the assumption that if the technology works, the experience will follow. It does not work that way. At enterprise scale, the human layer is not a polish pass you do at the end. It is the whole game. And right now, too many teams across the tech industry are shipping AI features without anyone in the room who is genuinely accountable for what it feels like to be on the receiving end of them. If we care about AI adoption, don’t forget the humans we are designing for. — Arin Bhowmick, SAP
Leaders worry about AI replacing jobs. They should also worry about AI replacing perspective. If the same institutions train the models, own the infrastructure, and define the use cases, we risk creating a future with extraordinary intelligence but limited imagination. — Hala Hanna, MIT Solve
The AI risk I think leaders underestimate is the impact of deepfakes on trust and brand relationships. A convincing piece of AI-generated content can spread globally in hours and create real reputational or commercial damage before legal processes can catch up. As AI becomes more powerful, protecting authenticity and intellectual property will become just as important as finding new ways to use the technology. — Ben Jeffries, Influencer
The most underestimated AI risk is the adoption gap. Most ROI models assume that once a tool is built or purchased, people will use it consistently and correctly. They rarely price the work required to change behavior: workflow redesign, training, incentives, governance, and accountability. In many programs, that work is as consequential as the technology itself. When adoption is not funded, owned, and measured, the value never becomes visible. The initiative either stalls or remains stuck in pilot because leadership never sees enough evidence to move it into production. — Andra Vaduva, SafeSpace
The risk is that we kill company culture, the slippery quality that makes people want to give their all and over-commit to each other and the mission. In most large companies without a really acute product moat, culture eats strategy for breakfast. The leaders that forget that (and I think a lot of them are right now) will pay a heavy price. To be clear, being on the front foot of AI transformation and having a strong company culture aren’t antithetical, but the balance needs to be treated as a priority, not an afterthought — Neil Barrie, 21st Century Brand
The AI risk leaders underestimate is the erosion of critical thinking. I’ve seen work that looks impressive on the surface, but when you ask one question deeper, the reasoning falls apart because the thinking was outsourced to AI. The future advantage won’t come from using AI; everyone will do that. It will come from judgment, curiosity, and knowing when the AI is wrong. — Emily Kortlang, Yerba Madre
There is still a tremendous amount of fear among employees about how AI will impact their jobs and the value they create for the business. We are still in the nascent stages of AI and understanding what it will mean for workers, and leaders are underestimating the amount of uncertainty, fear, and trepidation their team might have in adopting it. Having real conversations about what we know, what we don’t know, and what to expect will help guide us all through this massive cultural shift. — Regan Parker, ShiftKey
Leaders often underestimate how quickly AI can reinforce misalignment across the business. AI is incredibly powerful at generating insights, but if teams are operating in silos or measuring success differently, it can actually amplify disconnected decisions rather than unify teams. When implementing AI, I’ve seen how important it is to align teams from the very start—because without shared context and goals, even the best data won’t translate into better outcomes. — Melissa Puls, Ivanti
One AI risk leaders underestimate is over-controlling it. When governance feels slow, abstract, or disconnected from the business, people find ways around it. That creates more risk, not less. Good governance should help teams move faster on the right things, while being clear about where the real boundaries are. — Todd James, Aurora Insights LLC
One of the biggest AI risks leaders underestimate is assuming technology alone will drive transformation. AI can accelerate efficiency and insight, but without human judgment, creativity, and collaboration, its impact is limited. The organizations seeing the greatest results are embedding AI into their culture, workflows, and decision-making—not simply adding it to their technology stack. — Susan Watts, SPACECRAFT
Leaders are underestimating accuracy. Volume and speed do not imply accuracy. Just because something is now easier does not make it right. — Alex Triplett, You.com