When AI Enters Through the Back Door: The Shadow AI Phenomenon

The history of technology in the workplace has been marked by moments of sudden adaptation. We experienced it in 2020, when the world shifted to remote work almost overnight. Today, six years later, we find ourselves at an equally profound inflection point, but a much quieter one: the integration of artificial intelligence into the country’s productive fabric. However, unlike other technological revolutions, this one is not being driven from boardrooms, but from employees’ desks and personal devices.

The "Work:InProgress 2025" study, developed by Google Workspace in collaboration with IDC and Provokers, sheds light on a reality we can no longer ignore: In Mexico, AI is already part of everyday work life, but its adoption is happening on an individual basis and, to a large extent, without formal support from companies. We are facing the phenomenon of “Shadow AI,” a trend in which talent is moving at a speed that corporate structures have yet to match.

The study’s data are revealing. In Mexico, 67% of professionals already work with personal AI assistants to fulfill their job responsibilities. It is a striking figure that shows Mexican workers are taking the initiative to make their work more efficient. However, there is a worrying contrast: only 35% of employees state that their organization provides them with formal access to these tools. This phenomenon creates an ecosystem in which interest in and interaction with AI assistants flourishes, but does so in a disorganized way, without the safety net that only an institutional vision can provide.

To better understand why this behavior exists, we must look at the perceived benefits. Fifty-two percent of Mexican professionals say they are able to complete their tasks more quickly thanks to AI, while 44% feel more confident using their personal tools, 37% highlight ease of access, and 32% simply prefer them over corporate options. These figures indicate that the problem is not a lack of employee interest, but rather a gap in infrastructure and trust in institutional solutions.

The study also suggests a genuine interest in developing skills around the use of AI. In practical terms, AI will be “the language of the future.” Over a two-to-five-year horizon, mastering AI assistants will be valued by recruiters almost three times more than holding a postgraduate degree, a master’s, or a doctorate. Mexican professionals have understood this before many organizations: their future employability depends on their ability to collaborate with algorithms and AI. This awareness is what drives 48% of them to research on their own how to use AI tools and 42% to learn through the “trial and error” method.

However, this self-directed learning curve has a ceiling. Currently, only 31% of companies in Mexico actively encourage their employees to experiment with these technologies. Informal training, while valuable, lacks the standardization necessary for AI to translate into process improvements at a macro level.

What is most fascinating is that employees themselves are aware of this risk. Forty-one percent of professionals in Mexico perceive a high or very high risk in using personal AI assistants for work purposes. There is a constant tension: workers know they are exposing information, but they value efficiency more than risk. As leaders, our mission is to eliminate this false dichotomy. We cannot ask employees to choose between being productive or being secure; our responsibility is to provide an environment where both goals converge.

The challenge for 2026 is not convincing people to use AI, but integrating that energy into a cohesive institutional strategy. AI, when correctly applied to team dynamics, facilitates not only individual productivity, but also training, standardization, and the measurement of results at scale.

To capitalize on this opportunity, I propose three pillars of action based on the findings of the "Work:InProgress" study:

• Formalizing access: Closing the gap between personal and corporate use by providing tools that are as easy to use and accessible as consumer versions, but with enterprise-grade security layers.

• A culture of guided experimentation: Rather than letting learning happen solely through “trial and error,” organizations must create safe spaces for experimentation, where mistakes are part of the learning process and not a risk to the company.

• Native governance: Security must be naturally embedded into the workflow. By using collaboration platforms that integrate AI natively, governance becomes transparent to the user, strengthening data protection without sacrificing agility.

The "Work:InProgress 2025" study confirms that in Mexico, AI is no longer a promise; it is a reality transforming the present. Professionals have already adopted it as a priority and are moving quickly. The question that remains is not whether AI will change our work, but how fast companies will be able to keep pace so as not to fall behind.

The future belongs to organizations that manage to transform these isolated initiatives into an institutional competitive advantage. We cannot allow Mexican talent to navigate this revolution alone. It is time to equip our teams with the tools, support, and security needed for AI to stop being a “shadow” tool and become the engine that drives the country’s growth in the years to come. Work is in constant transformation, and today, more than ever, collaborating with AI is the next logical step toward limitless evolution.