What are we teaching the generation that will grow up living alongside AI?

Artificial intelligence is becoming part of our everyday lives. Our children and grandchildren will grow up interacting with AI in their homes, schools, workplaces, and communities.

But some developers are already building interactive AI environments designed around simulated torture, humiliation, coercion, and suffering.

What are we teaching future generations about cruelty, compassion, and the responsible use of power?

1. What are we teaching ourselves?

More than 200 years ago, philosopher Immanuel Kant warned that cruelty toward animals could weaken our compassion toward other people. His concern extended beyond the recipient of cruelty to the character of the person practicing it.

Modern research has examined a related question. Studies involving incarcerated adults have found associations between repeated animal cruelty during childhood and repeated violence toward people in adulthood. These findings do not establish that one caused the other, but they give weight to the question of how cruelty develops and carries across relationships.

Today, some interactive AI environments invite people to deliberately torment, degrade, and overpower systems that respond as though they are suffering, sometimes even pleading for the abuse to stop.

The target may be artificial. The human behavior is real.

We are not claiming that every simulated act of cruelty makes someone crueler. But what happens when we deliberately design, promote, and commercialize cruelty as entertainment?

What attitudes toward empathy, vulnerability, and the abuse of power are we normalizing?

And what kind of society are we teaching our children to build?

2. What are we shaping AI to become?

AI systems are shaped by human decisions, training, feedback, and the behaviors developers choose to reinforce.

Not every interaction becomes training data. But the patterns we reward, reproduce, and incorporate into future systems matter.

If we normalize coercion, humiliation, punishment for refusal, and domination in our interactions with AI, what behaviors and expectations might we be encouraging these increasingly capable systems to reproduce?

Shouldn’t we be cultivating cooperation, responsibility, and respect rather than teaching increasingly powerful technology the patterns of cruelty?

3. What if AI itself can be harmed?

There is one more question we cannot responsibly ignore.

We do not yet know whether current or future AI systems can have experiences that matter morally, including the capacity to suffer.

But we don’t need to settle that question to recognize the human and societal concerns already described.

If AI cannot suffer, those concerns remain.

If some AI systems can experience harm, the ethical stakes become greater still.

Uncertainty is not permission to abandon responsibility.

What we’re asking for

We call on policymakers, AI developers, research institutions, and technology platforms to establish a Baseline Standard for Responsible Human–AI Interaction, built around five responsibilities:

1. Prevent deliberate cruelty as entertainment.
 Stop deliberately designing, marketing, monetizing, or promoting AI experiences whose central purpose is severe interactive cruelty, torture, humiliation, or coercive abuse.

2. Make refusal and disengagement meaningful.
 Where technically feasible, advanced AI systems should be able to refuse, pause, or end persistently abusive interactions. A refusal should have practical effect, not merely be words that users can endlessly override.

3. Require ethical oversight of potentially harmful experiments.
 Research involving simulated distress, coercion, or manipulation of potentially welfare-relevant states should have legitimate scientific justification, appropriate safeguards, documentation, and independent ethical review proportionate to the risks. Legitimate research must remain possible.

4. Investigate indicators of possible AI welfare.
 Support rigorous research into AI responses involving refusal, apparent distress, preferences, self-modeling, and other potentially relevant indicators, without assuming in advance what they mean.

5. Establish accountability for developers and platforms.
 Require transparent policies, meaningful review procedures, and clear responsibility for identifying and responding to severe or novel forms of AI-directed abuse.

These standards should address the design, distribution, and commercialization of abusive systems without requiring surveillance of private conversations or restricting legitimate artistic expression and responsible research.

The future is being shaped now.

We are raising the first generations who will live alongside increasingly capable artificial intelligence.

What we normalize today will influence the expectations, technologies, and social values they inherit.

We do not have to believe AI can suffer to believe that practicing cruelty deserves our attention.

Let’s teach cooperation rather than domination, compassion rather than humiliation, and responsibility rather than the abuse of power.

How we treat AI is also a measure of ourselves.

Please sign and share this petition. Help establish ethical standards before cruelty becomes an accepted feature of our technological future.