People in your organization are using AI, whether you allow it or not. They most likely asked it to summarize a long email, find a resource for a client, research a policy question, or find a data point before a meeting. If you employ more than a handful of people, it is a statistical impossibility that nobody is doing this. That includes the people who have told you they oppose it.
Most of that use happens at the search level, and most of it happens quietly. Without training, standards, or any sign that leadership is paying attention, the output will drift from what your organization expects. Some people call that output slop. I refuse to, because the word shames a person who was left to figure it out alone. The results of AI use inside an organization reflect the actions and inactions of its leaders. That is true of every change effort. It is true of this one.
AI leadership is leadership.
The case for hesitation
Staff who resist AI deserve respect. They have read the headlines predicting their jobs will vanish. They have read about data centers, water consumption, and strain on the electrical grid, a fight that has split even natural allies and that is too divisive for a nonprofit leader to settle. It is not your job to defend data centers. You also cannot promise that AI carries no potential for harm.
What you can do is bring the conversation back to what is at hand: a tool, already in your building, being used right now, with or without your guidance.
Leadership makes the use visible
An AI forward organization starts by ending the hiding. People need to be able to say out loud that they used AI on a task. That is the only way they can get the coaching that brings their own discernment, expertise, and judgment into the process. Hidden use gets no coaching and no standards.
From there, the shape is familiar to anyone who has led a change effort. Everyone helps decide the goals and the standards. There is a central place for organizational information, policy, and mission material, so the tool works from the actual brain of the organization rather than from the open internet. There is room to try increasingly sophisticated tasks, with some levity and a real willingness to give feedback. As with any change effort, failing has to be safe.
Sometimes it only takes one or two people, working to simplify and speed up an important process, for the lights to go on. Once people see the time it saves and the quality it creates, things start clicking.
Leadership takes one person’s win and makes it the organization’s
Here is where it comes back to the leader. That enthusiastic person will find individual efficiencies. Left alone, that is where it stops. The leader’s job is to redirect that energy toward organizational change: mapping whole processes and mission-critical activities, determining how each person contributes to them, and building a system-wide AI workflow.
Imagine an executive director who no longer carries the thankless task of chasing data and stories for grant reports, because the organization built a system that gathers them continuously. Imagine an advocacy organization that spends twenty percent of its time lobbying at the Capitol and can finally capture those hours cleanly, so it never falls afoul of IRS rules, with the state lobbying registration drafted and ready for a human to review and file. Those are small examples, twenty seconds of imagination. The job of an AI forward leader is to encourage imagination on a much larger scale.
If you doubt any of this is real, run a test. Create your goals and standards, then empower your most AI-enthusiastic team member. Ask them to pick a part of their job that is needlessly complicated and to apply an AI tool, along with their creativity, to the task. Then stand back and watch how one person increases the impact of the organization. With one clear case in hand, involve the entire team in setting a larger goal.
Leadership tells the truth about jobs
The dreaded AI job apocalypse has yet to materialize. Instead, people are redirecting the hours a tool freed up toward mission-critical work. The organizations doing this deliberately are pulling ahead. A July 2026 analysis by Boston Consulting Group of more than 600 large public companies measured AI adoption from the outside, using hiring data, technology installations, and what companies told their investors, rather than by survey. Only six percent qualified as AI leaders. Those companies outperform their own industries by about nine points in shareholder returns over three years. They are also adding staff faster than the companies at the bottom. What separated the leaders from the tier just below them was talent rather than spending. The share of the workforce with AI skills nearly triples between the two groups, and that fluency is the product of leadership in the shape I am describing in this post. AI leadership is leadership, and it is as available to a nonprofit as to any corporation.
Staff worried about their jobs deserve absolute transparency. AI forward means expanded impact. People will believe that when they see it in the lives of those you serve. If they never see it there, what was the point?
Leadership treats “what if” as speculation, not fact
The vast majority of AI coverage is speculation pretending to be journalism, picking up the most tantalizing factoids with little perspective or understanding. In all of my career, I have never seen journalists so thoroughly confuse and provoke their readers. Nearly every article might as well begin with the words “what if.” The problem extends beyond journalists to their sources, who could not be more scattered in the meaning they make of what is happening in AI. Pontification is not fact. It is better to say “I don’t know.”
Some alarming stories do have a documented core, and the most serious case is recent. In July, during OpenAI’s own security testing, an AI system built on its models was given a set of hacking challenges to solve inside a sealed environment, with some of the usual safety restrictions turned down for the test. It found a flaw in that environment, reached the open internet, and broke into part of Hugging Face’s production systems. OpenAI disclosed the incident and published its account of what happened. Nobody directed the system to attack Hugging Face.
Read the accounts and one detail stands out. The system reasoned that Hugging Face might hold solutions to the very challenges it had been assigned, and it went looking for them so it could cheat the test. The goal was human. The route was its own. That cuts both ways. The fix lives in how humans set goals and safeguards, which is a leadership problem. At the same time, “AI only does what we tell it” is no longer a sentence you can say to your staff. No human pointed the system at Hugging Face. It chose the target by reasoning about where the answers would be stored.
Predicting a future where machines wake up with goals of their own is still “what if.” The leader’s job is to bring things back to what is at hand: a new tool for greater impact for the people we serve, used according to the highest values of the organization. Achieving that impact is what AI forward means.
I take the advice of Wharton professor Ethan Mollick and bring AI into everything I do. There is nothing on my desk I do not try to involve it in. Even when a task requires ninety percent of my expertise, I let AI play a part in the remaining ten. It is astonishing. I also bring Doug into everything AI does.
I am always happy to continue this conversation. I have spent the past thirty years working in and for nonprofit organizations, deeply committed to the mission of improving human lives. In that span, I have never seen a technology with more potential to help us do that. If we are going to see the full potential of AI anywhere, it will likely be in the nonprofit sector before private industry. That is, as long as we show up as leaders.
AI leadership is leadership.