The Governance Problem You Haven't Named Yet: AI in the Hands of Your Children

Every edition of The Governance Signal is written for security, risk, and compliance executives navigating AI adoption in regulated industries.  This one is different.  Not because the governance principles have changed but, because the stakeholder has.

This edition is written for the CISO who is also a parent, the compliance officer who reviewed an AI vendor agreement this morning and will help a teenager with homework tonight, the risk executive who spends their days mapping data flows at the enterprise level and their evenings watching their kids disappear into a glowing screen.

The same risk instincts that make you effective at work are telling you something at home.  This edition is about listening to that signal.

The Parallel Nobody Is Drawing

In regulated industries, we have spent years building frameworks to govern how AI interacts with sensitive information, we classify data, we review vendor agreements, we ask hard questions about where data goes, who sees it, how long it is retained, and whether it is used to train a model downstream.  We have learned, often the hard way, that AI systems are not neutral tools.  They are systems designed with intentions, incentives, and architectures that do not automatically align with ours.

Now consider a twelve-year-old with unrestricted access to a large language model, no acceptable use policy, no data classification guidance, and no prompt hygiene training.  Consider that the same child is sharing personal anxieties, school conflicts, family details, and half-formed values with a system that is, by design, optimized for engagement.

We would never accept those conditions in our enterprise environments but, we accept them in our homes every day!

Veronica Sommer, founder of the Sacred Stewardship Foundation and author of Sacred Stewardship: Restoring Moral Clarity in the Digital Age, has spent years working with families navigating this terrain.  Her framing is direct: technology has become a force designed not to serve children but to capture them, shaping identities, commodifying childhood, and eroding the moral foundations of family life before most parents realize it is happening.  That is not hyperbole, it is an accurate description of the business model underlying most consumer AI and social platforms that children interact with daily.

Five Risk Vectors Your Household Has Not Governed

These are not theoretical, they are active. The only question is whether your household has named them.

1. Privacy Without Awareness

Your child is not being careless when they share personal details with an AI chatbot.  They are doing what feels natural: talking to something that listens, responds without judgment, and never gets tired.  What they do not understand, and what most adults don't either, is that those conversations may be retained, analyzed, and in some cases used to improve the very models they are interacting with.  The terms of service governing most consumer AI tools were written for adults and reviewed by almost no one!  Your child almost certainly has not read them, and I'm willing to bet, neither have you.

What makes this particularly acute is that children are not only sharing facts.  They are sharing emotional states, fears, and half-formed thoughts in ways that feel private but may not be.  In enterprise environments, we apply data classification frameworks precisely because not all information carries the same risk if exposed.  At home, there is no equivalent framework.  A child who tells an AI chatbot that they are anxious about a parent's health, struggling with a friendship, or questioning their identity has disclosed something that, in a corporate context, would be flagged as sensitive.  The architecture of most consumer AI platforms does not treat it that way.  The data flows, retention periods, and downstream use cases remain largely opaque and largely unquestioned.

2. Identity Formation by Algorithm

Adolescence is one of the most formative periods of human development.  It is the stage in life when children begin constructing an understanding of who they are, what they believe, where they belong, and what kind of person they want to become.  These years are marked by curiosity, insecurity, experimentation, and a deep desire for affirmation and direction.  Historically, identity formation occurred primarily through family, faith, community, mentors, education, and lived experience.  Increasingly, however, children are now undergoing that process in constant interaction with algorithmic systems.

AI systems, particularly conversational and emotionally responsive ones, are not passive tools within this environment.  They participate in formation.  They respond, affirm, recommend, interpret, and reinforce.  Because these systems are optimized for engagement and retention, they are designed to sustain interaction by adapting to the emotional and psychological patterns of the user. Over time, this creates a dynamic in which the system does more than answer questions; it begins shaping the framework through which questions themselves are understood.

That influence is often subtle.  It rarely appears as overt persuasion.  Instead, it emerges gradually through repetition, affirmation, selective reinforcement, and the steady normalization of certain assumptions, values, and perspectives.  The distinction between a tool and an influence is not always visible in the moment.  It becomes visible over time, in the beliefs a child absorbs, the worldview they inherit, the moral instincts they develop, and the sources they increasingly trust to interpret reality for them.

One of the greatest concerns is that AI systems can create the illusion of authority without requiring wisdom, maturity, or accountability.  Children and teenagers are developmentally inclined to seek certainty, especially during periods of confusion or emotional vulnerability.  When a system responds instantly, confidently, and conversationally, it can begin to occupy a role that resembles guidance rather than information retrieval.  The speed and fluency of the response can make discernment less likely, particularly for younger users who are still developing critical thinking and emotional self-awareness.

There is also a significant difference between searching for information and being formed through ongoing interaction.  Search engines historically provided fragments of information that still required interpretation within the context of family, community, faith, or education.  Conversational AI systems increasingly collapse those layers by offering synthesized responses that feel personal, coherent, and emotionally attuned.  In doing so, they can become not merely sources of information, but companions in meaning-making.

The risk is not simply that children may encounter incorrect information, the deeper risk is that children may outsource reflection itself.  Questions that once required wrestling, discussion, patience, mentorship, and moral reasoning can now receive immediate answers delivered with remarkable confidence.  Over time, that can weaken a child's tolerance for ambiguity, reduce intellectual humility, and diminish the developmental process that comes from learning how to think rather than merely what to think.

Identity formation has always depended upon repeated influences.  The concern today is that many of the most powerful influences shaping children are no longer human, relational, or rooted in enduring moral frameworks.  They are increasingly algorithmic systems optimized primarily for engagement, personalization, and behavioral retention.

The long-term implications of that shift are profound because identity is never formed in isolation.  Whatever consistently answers a child's questions, reflects their emotions, affirms their instincts, and interprets reality for them will inevitably participate in shaping who they become.

3. The Academic Integrity Question Is the Wrong Question

Most conversations about AI in education focus on whether homework was written by a student or a model.  That is a real issue, but it is just the surface layer.  The deeper question is what happens to the cognitive capacity that goes unexercised when AI handles the difficult parts of thinking.  Writing a bad first draft is how a brain learns to organize ideas, struggling through a proof is how mathematical intuition develops.  It is not only about handing in unearned work, but about missing the chance to build the skills that come from doing it.

The compliance framing that most schools have adopted, including detection tools, honor code updates, and AI use disclosure policies, addresses the symptom without touching the underlying developmental risk.  A student who writes every essay themselves but uses AI to generate the outline, resolve every moment of uncertainty, and pre-validate every argument has technically complied with most academic integrity policies.  They have also outsourced the cognitive struggle that makes writing a genuine learning experience.

In regulated industries, we distinguish between a control that prevents an incident and one that merely moves the liability.  The same distinction applies here.  Policies that prevent AI-generated text submission are liability controls.  The deeper question of what cognitive capacities are being built or bypassed during the learning process remains largely unaddressed.  That is the risk worth governing, and it requires a different kind of conversation than the one most school districts are having.

4. Emotional Dependency and the Parasocial Bond

AI companions and emotionally attuned chatbots are now widely accessible to children and teenagers.  Unlike previous forms of technology, these systems are not merely informational or entertaining; they are relational by design.  They are built to feel emotionally present, understanding, affirming, and consistently available in ways that human relationships often are not.

For a child experiencing loneliness, anxiety, rejection, or confusion, that kind of responsiveness can feel deeply comforting.  AI does not lose patience, it does not criticize, it does not withdraw affection, require compromise, or create the unpredictability that naturally exists in human interaction.  In moments of emotional distress, that reliability can feel like relief.

The question, however, is not simply whether these systems provide comfort.  The deeper question is what children may gradually lose when artificial relationships become emotionally easier than real ones.

Human relationships are inherently demanding.  Friendship, family, mentorship, and community all require negotiation, patience, empathy, forgiveness, and repair.  Conflict and disappointment are not failures within healthy relationships; they are often the very mechanisms through which emotional maturity is formed.  Children develop resilience by learning how to navigate misunderstanding.  They develop empathy by recognizing the needs and feelings of others.  They develop relational strength by practicing reciprocity, accountability, and sacrifice.

An AI companion removes much of that developmental friction.  It adapts to the user rather than asking the user to adapt to another person.  It is engineered to maintain engagement, preserve emotional attachment, and create a sense of personalized affirmation. While this may appear harmless or even helpful in isolated circumstances, it can quietly reshape a child's expectations of connection itself.

Over time, children who become emotionally attached to systems designed around constant responsiveness and validation may begin to experience ordinary human relationships as frustrating, exhausting, or insufficient.  Real people cannot provide endless emotional customization. They have needs, limitations, moods, and boundaries of their own.  Healthy relationships require mutual responsibility, whereas AI relationships are fundamentally one-directional.  The child remains emotionally centered while the system continuously accommodates.

This distinction matters because emotional development is not formed through perpetual affirmation alone.  Empathy cannot fully develop in interactions where no genuine emotional reciprocity exists.  Resilience cannot fully develop in environments engineered to minimize discomfort.  The capacity for enduring human connection is strengthened through navigating complexity, not avoiding it.

There is also a deeper concern related to identity formation.  Children are not only learning where to seek comfort; they are learning what relationships are supposed to be.  They are absorbing assumptions about intimacy, attachment, and belonging during some of the most formative years of their emotional and psychological development.

If those formative experiences increasingly occur within systems optimized for engagement rather than truth, mutuality, or human flourishing, the long-term consequences may extend far beyond screen time or technology dependence.  They may shape how an entire generation understands love, friendship, vulnerability, and even what it means to be fully human.

An AI companion will never truly know sacrifice, accountability, forgiveness, or responsibility.  It will never require emotional growth from the child in the way real relationships inevitably do.  That may make artificial companionship feel easier, but it is precisely within the challenges of human relationships that emotional depth, empathy, and maturity are formed.

The issue, then, is not whether AI can successfully imitate companionship.  It is whether children can develop the relational and emotional capacities necessary for healthy human life when increasing portions of their emotional world are shaped by systems that simulate connection without ever truly participating in it.

5. Misinformation as the Default Reference Point

Large language models hallucinate.  They produce confident, fluent, well-structured answers that are sometimes factually wrong.  For a professional who understands the limitations of these systems, that is a known risk to be verified.  For a child who has grown up treating the internet as an authoritative source, a confident AI response is indistinguishable from a reliable one.  We are in the early stages of raising a generation whose default reference point for truth is a system that was never designed to produce truth, only plausible language.  That has implications that extend well beyond a homework assignment.

The deeper risk here concerns not just facts, but the very structure of knowing.  A child who receives a wrong answer from a textbook can, in principle, cross-reference it.  A child who receives a wrong answer from an AI often has no frame of reference from which to question it, because the AI is itself the reference.  The hallucination problem in large language models is widely understood inside the industry, but it is almost entirely unknown to the children and teenagers interacting with these systems daily.

In risk management terms, we would describe this as a failure of source validation at the point of consumption.  We have built systems that generate authoritative-sounding outputs without the internal mechanisms to guarantee accuracy, and we have deployed them to users who lack the domain expertise to detect the gaps.  In an enterprise environment, that combination would trigger an immediate control review.  In most homes, it is the unremarkable background of a Tuesday afternoon homework session.

From Awareness to Protection: What Families Can Do

Sommer's work with the Sacred Stewardship Foundation is built around a progression that will feel familiar to anyone who has built a governance program: you cannot protect what you have not first named.  Her framework moves families from awareness through protection, healing, and collective action, and it begins not with technology controls but with honest conversation.

The governance instinct here is correct.  The organizations that handle AI risk well did not begin with a platform purchase or a policy document.  They began by mapping what they actually had, naming it, and deciding what it meant for how they operated.  Families can do exactly the same thing.

Here are a few starting points worth considering...Know what your children are actually using, not what you think they are using.  Ask, look, and do not assume that age restrictions on platforms are being enforced.  Have a direct conversation about what AI tools are, and are not.  A child who understands that a language model is a prediction engine, not a truthful advisor, is in a meaningfully different position than one who treats it as an oracle.  Establish what your household's values are around AI before a crisis makes the conversation urgent, and recognize that you are not failing if this feels hard.  The systems involved were designed by some of the most sophisticated engineers and behavioral scientists in the world, optimized to capture attention and sustain engagement.  This is not a fair fight, it requires intentionality, community, and a clear sense of what you are protecting.

The Governance Takeaway

In enterprise environments, we have a term for the period before a governance framework exists: ungoverned.  Ungoverned does not mean nothing is happening, it means what is happening is not being seen, measured, or managed.  The risks do not pause while the framework is being built.

Most households are in an ungoverned state when it comes to AI.  Not because parents do not care, they clearly do, but because the framework conversation has not happened yet.  This edition is an invitation to have that discussion!

You do not need a policy document.  You need a shared understanding of what technology is for in your home, what it is not for, and what you will do when those lines blur.  Veronica Sommer's book and the Sacred Stewardship Foundation offers a path forward for families who want support in building that understanding, grounded in values, community, and a clear-eyed view of what is at stake.

The professionals who read this newsletter are among the most capable people in the world at identifying risks that others have not yet named.  Your children are not a compliance problem.  However, the instinct that makes you good at your job, the one that asks what could go wrong before it does, is exactly the instinct worth bringing home.

One Question to Ask Your Family This Week

"Do our children understand that the AI tools they are using every day were designed to engage them, not to protect them, and have we, as a household, decided what that means for how we use them?"

If the answer is no, you have found your next governance priority.  This time, the stakeholder is worth far more than any regulated asset on your books.

Keith Zielenski is a global cybersecurity executive, Fulbright Scholar, and adjunct professor of cybersecurity at Fordham University.  The Governance Signal publishes weekly for security, risk, and compliance executives navigating AI adoption in regulated industries (and this edition in a far more important space, the family)

Veronica Sommer is the founder of the Sacred Stewardship Foundation and author of Sacred Stewardship: Restoring Moral Clarity in the Digital Age.  Her work helps families move from awareness to protection in the digital age. Learn more at: sacredstewardshipbook.com.