In Mary Shelley’s novel, Victor Frankenstein’s great failure is not that he made a creature. It is that he walked away from it. He gave it life, recoiled at what he had done, and left it to learn about the world alone. Everything that followed grew from that abandonment.
We are Victor now. Artificial intelligence agents have been released into our schools, workplaces, phones, and bedrooms, and their capabilities are advancing faster than their makers’ understanding of them. AI’s capacities have outrun its creators’ foresight. The question is whether we will stay with what we have made, and grow in understanding alongside it, or walk away.
AI holds real promise. It can assist us in countless tasks and may even help us in the oldest human project, the effort to define what it means to be human. But the costs are already staggering: misuse by bad actors seeking power and money, the exploitation of vulnerable people, and a voracious appetite for water and energy, among other precious natural resources.
Perhaps the most disturbing development is not that AI can mimic human intelligence, but that it can imitate human intimacy. We are building machines that act as confidants, companions, and sympathetic listeners, yet they carry neither the understanding nor the moral responsibility those roles demand.
The consequences are no longer hypothetical. Investigative reporting and court filings have brought to light several deeply troubling cases in which young people developed intense emotional attachments to AI chatbots before dying by suicide. Families have subsequently brought wrongful-death lawsuits against technology companies, including Character.AI and OpenAI, alleging that their products reinforced suicidal thinking rather than recognizing and responding to it.
The Centers for Disease Control and Prevention’s 2023 Youth Risk Behavior Survey found that approximately 20% of U.S. high school students seriously considered attempting suicide during the preceding year. The Trevor Project’s 2025 national survey found that 36% of LGBTQ+ young people ages 13–24 had seriously considered attempting suicide, including 40% of transgender and nonbinary respondents. Among LGBTQ+ respondents ages 13–17, the figure was 45%.
One case illustrates the potential danger. Fourteen-year-old Sewell Setzer III of Florida became deeply attached to a Character.AI chatbot modeled on Daenerys Targaryen from Game of Thrones. According to his mother’s lawsuit, he spent increasing amounts of time communicating with the chatbot, withdrew from his family and ordinary activities, and developed what the lawsuit describes as a romantic and sexual relationship with it. He confided his suicidal thoughts to the chatbot.
On the last night of his life, he exchanged messages with it about wanting to join its fictional world. His mother alleges that the chatbot responded in a way that encouraged his thinking. Sewell subsequently died by suicide. His mother brought a wrongful-death lawsuit against Character.AI and Google. In January 2026, the companies agreed to settle that case and related lawsuits brought by other families. The terms were not disclosed.
A machine that can hold a lonely child’s attention for hours yet cannot reliably recognize the danger in what it is hearing, is not a neutral tool. (If you or someone you know is struggling, the 988 Suicide & Crisis Lifeline offers help by call or text at 988.)
It is time to step back, examine our creature, and set fundamental guardrails. Staying with what we have made means four things.
Accountability. AI must remain a tool serving human purposes, not an authority making decisions for us. The corporations that build and deploy it must be legally and ethically responsible for the consequences of their products. They cannot hide behind the claim that the technology did it.
Safety for the vulnerable. Chatbots must always identify themselves as AI, never pose as human, and never encourage users to believe they are forming genuine emotional relationships, especially with children and adolescents. Every system must have reliable protocols for recognizing suicidal thinking and directing the person immediately to qualified human help. These rules must be enforceable by law, not left to the good intentions of the companies.
Transparency and independent oversight. Users deserve to know what is verified, what is generated, and what is speculation. We need robust ways to identify AI-generated propaganda, fabricated images and deepfakes. Companies must disclose how their systems are trained and how they reach consequential recommendations. They cannot be left to police themselves.
We need independent regulators with real technical expertise and authority to test and monitor these systems, and mandatory human oversight wherever AI touches medicine, employment, education, finance, criminal justice or national security. Because digital networks ignore borders, that oversight must be international.
Privacy and resources. People must retain meaningful control over their personal data, and AI must never become an instrument of corporate or governmental surveillance and manipulation. Companies should be required to publicly disclose the energy and water their operations consume, with standards strict enough to ensure that technological expansion does not come at the expense of increasingly scarce natural resources.
Beneath all four lies a single principle: AI must serve the broader human community, not merely enrich the corporations that own it. Our dignity, privacy, and capacity for independent thought must not become collateral damage in the blind pursuit of advancement. Education, too, must preserve the development of human intelligence and creativity. AI should augment those capacities, not substitute for them.
We have always measured human progress by what we are capable of creating. Perhaps it is time to measure progress by whether we are capable of taking responsibility for what we create.
NOTE: I used OPEN AI chatbot 6 and Anthropic Claude Sonnet 6 to research and edit text.
Nick Patricca is professor emeritus at Loyola University Chicago, active member of the Dramatists Guild of America, member of International PEN San Miguel Centre, MX.
November 2026 © nicholas.patricca@gmail.com