MIT Technology Review op-ed: AI safety laws miss everyday harms

MIT Technology Review has published an opinion essay by Sam Hiner and Emma Leiken. Its argument: technology policy treats success as preventing the worst thing a product might do, and that bar is low enough for companies to clear while still building products that leave young people dependent on simulated friendships or trapped in feeds designed to keep them scrolling.

The authors open with two examples of what passes for success: an AI company promising its chatbot will not encourage suicide, and a social media platform making it harder for children to buy fentanyl. These are important protections that should have existed all along, they write, but meeting them says little about everyday damage. A teenager with an unhealthy attachment to an AI companion, a student who has a chatbot write every paper, or a young person measuring their life against an envy-inducing feed will rarely appear in a lawsuit or a newspaper. A product can meet every narrow safety requirement and still leave the culture worse off.

The authors say a safety-centred definition of success is politically convenient for the technology industry and that companies actively promote it. Preventing spectacular disasters, they argue, means fewer lawsuits and less pressure for serious regulation, whereas confronting everyday dependence would threaten the bottom line far more. It might mean asking companies to add friction to their products or to limit engagement and features that lead to negative outcomes, such as addictive attention-hacking algorithms.

Their main example is legislation. This year, they write, Google backed bills to regulate AI chatbots in several states. The bills mandated basic protections against sexual content and promotion of self-harm, but with broad carve-outs that could exempt some of the most popular services. Passing them let legislators claim a safety win while efforts to promote stronger laws, designed to stop children from becoming emotionally dependent on AI, failed. The authors say this playbook worked in states like Nebraska, producing legislation heavily influenced by industry at the expense of bills that would have held companies more accountable.

They compare this with social media. Globally, people who use the internet spend an average of more than two hours a day on social media platforms, which the authors describe as time quietly taken from ordinary life. They write that the cultural engineering of these algorithms can contribute to violence on a devastating scale, and that Meta's platforms have been cited for contributing to the Rohingya genocide in Myanmar, anti-Muslim riots in Sri Lanka, and alleged calls for violence during Ethiopia's Tigray war. The same incentives, they say, operate in the feeds that shape young Americans' understanding of politics and their peers.

AI, the authors argue, extends this logic into parts of life that social media could influence only from a distance, by offering to write our thoughts and stand in for the people we might otherwise turn to. They cite a figure that a third of teens now choose to discuss serious matters with AI companions instead of other people. They also cite MIT researchers who found that young people using ChatGPT to write essays showed weaker brain connectivity during the task, with evidence that "cognitive debt" persisted after the task was complete. There is plenty still to learn about long-term effects, the authors say, but already reason to ask whether technology marketed as making us more capable is eroding the habits and relationships that build capability. Judgment, they write, is built through effort and friction, and we learn to live with others by confronting the fact that they have needs and opinions of their own.

The essay ends with demands. Safety should be a minimum, and then people should demand considerably more. Technology companies and school districts should have to show beyond a reasonable doubt that AI improves learning and critical thinking before making it a fixture in classrooms. Companies should be prohibited from designing companions that pretend to be conscious or profess love to users, and they should face independent scrutiny of how their systems affect relationships and people's ability to think for themselves.

Hiner is cofounder and executive director of the Young People's Alliance, where he coordinated a coalition of advocates to pass laws regulating AI companions in eight states in 2026, such as California's SB1119. Leiken is executive director of the Responsible Technology Youth Power Fund at Hopelab, which has raised more than $9 million for youth-led tech accountability work. She teaches technology policy at the Columbia School of International and Public Affairs and was previously TikTok's global head of youth safety and well-being policy.

Key facts

• Hiner and Leiken argue that tech policy defines success as avoiding catastrophic harms, a bar companies can clear while still building products that foster dependence.
• They say Google backed chatbot bills in several states this year that mandated basic protections against sexual content and self-harm promotion but carried broad carve-outs; stronger anti-dependence bills failed, including in states like Nebraska.
• Cited figures: global internet users spend an average of more than two hours a day on social media, and a third of teens choose AI companions over other people for serious matters.
• They also cite MIT researchers who found weaker brain connectivity in young people using ChatGPT to write essays, with "cognitive debt" persisting after the task.
• Their proposals: schools and companies must show beyond a reasonable doubt that AI improves learning, companions must not pretend to be conscious or profess love, and systems should face independent scrutiny.

Why it matters

The essay challenges how AI and social media rules are judged. If a chatbot refusing to encourage suicide counts as a win, the authors say, lawmakers can declare victory while the slower harms continue: dependence on simulated friendships, outsourced writing and thinking, and feeds built to hold attention. Their point is that safety is a necessary floor, not the goal, and that the industry has an interest in keeping the debate on the floor.

Who it affects

Mostly young people: teens who confide in AI companions, students who rely on chatbots for assignments, and young users of social media feeds. It also concerns legislators who write chatbot bills, school districts deciding whether to make AI a fixture in classrooms, and technology companies that would face limits on engagement features and companion design under the authors' proposals.

How to use it

This is an opinion piece, so the practical content is the authors' list of demands. They want technology companies and school districts to show beyond a reasonable doubt that AI improves learning and critical thinking before adopting it in classrooms. They want a ban on companions that pretend to be conscious or profess love to users, and independent scrutiny of how AI systems affect relationships and people's ability to think for themselves. They also say addressing everyday harms might mean asking companies to add friction or limit engagement-driving features. Readers following chatbot legislation can use the essay as a lens: check whether a bill covers only basic protections or also dependence, and how broad its carve-outs are.

How solid is it

It is an argued opinion essay, not a study or news report, and the claims are the authors' own. The facts it leans on are given without sources in the text: the two-hours-a-day average, the one-third-of-teens figure, and the MIT finding come with no citation, journal, date or sample size. The Meta examples are phrased in the passive ("have been cited") with no citing party named. The account of Google-backed bills and the Nebraska outcome is the authors' characterisation; the essay names no bill titles or numbers for those and gives no response from Google or any other company. The authors also say plainly that much about long-term effects remains to be learned.

Risks and caveats

The authors are advocates in this field. Hiner's organisation coordinated a coalition that pushed AI companion laws in eight states in 2026, and Leiken runs a fund for youth-led tech accountability and previously worked at TikTok on youth safety policy. That background does not make the argument wrong, but readers should weigh it as advocacy. The essay concedes that evidence on long-term effects is incomplete, and the MIT result it cites is about brain connectivity during an essay-writing task, not proof of lasting harm in daily life. Proposals such as proving benefit "beyond a reasonable doubt" are stated as principles, without detail on how they would be enforced.

“The companies that have inserted their products into nearly every part of modern life have earned obligations far greater than avoiding the next terrible headline.”

— Sam Hiner and Emma Leiken, MIT Technology Review