WASHINGTON, October 20, 2026 – A teen launches an app “for five minutes.” Three hours have passed, and the feed has fed her content on body image, self-harm, or radicalization, and the algorithm knows exactly how to keep her scrolling. Is the company that engineered that algorithmic engine liable if that harms the kid?
It should be—just not in a broad way, but rather with a duty of care. Technology companies should be held liable for the algorithmic design features that deliberately and negligently target children in ways any reasonable company would understand to be a risk. Not every bad post, or every bad outcome—but the features that capitalize on the developing mind to generate maximum engagement.
The design problem
Personalized recommendations are not neutral conduits; they select what the user will see next, how often, and with what emotions. This is relevant because adolescents are still maturing their impulse control, sense of self, and reasoning skills, meaning an attention-maximizing algorithm may amplify content that the child didn’t seek out and lacks the tools to handle.
As legal scholars Tal Zarsky and Gideon Parchomovsky contend in their article “Targeting Children: Liability for Algorithmic Recommendations,” social media and other websites should be viewed as distinct from hosts. “Algorithmic targeting creates and manipulates content; it is not simply speech of the human author transmitted to a new user,” the authors write. The proposed solution from Zarsky and Parchomovsky: Civil liability for negligent design, incentivizing companies to limit algorithmic targeting toward children.
This is important. A platform should not be made liable merely for housing something that’s terrible. But a company should be examined when the product design predicts a vulnerable user and promotes harmful content that keeps a minor engrossed regardless of warnings.
Why “negligent design” fits
Critics fear that imposing liability would chill innovation or lead platforms to censor legal speech. True enough. But negligence law has a well-known standard: was the corporation negligent, taking into account dangers that were or should have been foreseeable to it?
Imposing a duty of care does not mean perfection. It means implementing reasonable age-appropriate defaults, minimizing addictive characteristics of design, providing appropriate parental controls, and protecting children when the recommendation system recognizes that a child shows signs of distress. As the authors of the SSRN article recognize, liability alone cannot be the solution; reform is needed in many other areas as well. Liability, however, can create new incentives in ways that voluntary commitment and heartfelt public apology have not.
Let’s face it. “The algorithm did it” is not much of a defense. The algorithm was designed, programmed, tuned, and made profitable by the very company that uses it. Suppose that an auto manufacturer produces brakes that inevitably fail whenever a teenager takes the wheel.
The hard part: causation
Here’s where the argument gets complicated. Proving that an injury was the result of a recommendation is complicated. Teenagers come into contact with dangerous content from different sources; families are not the same; mental health problems stem from different things. Evidence will be needed for the recommendation process, the platform’s awareness, and a link between the design process and the injury.
This is why a general strict-liability regime will be wrong. A negligence theory that allows the plaintiff to request discovery but requires him to prove unreasonable action will work better. The authors of the SSRN article recognize this evidentiary problem, along with First Amendment issues, but still argue for liability anyway.
What Congress should do
Congress doesn’t need to regulate every video, every comment, or every search. Congress needs to regulate the design decisions that turn children into the product.
In legal analysis published through the American University Law Review, scholars emphasize that federal legislation should establish a baseline duty of care for users under 18. This framework would prohibit default engagement-optimized feeds, ban manipulative design patterns that subvert age restrictions, mandate independent youth safety audits, and preserve evidence whenever a platform receives reports of harm—all while safeguarding encryption, user privacy, and protected speech.
The point here’s not to rid the internet of risk; this can’t be done. The point is to treat teens’ attention as a limited, not unlimited, resource.
The bottom line
The tech industry should be held legally liable in the case that their algorithmic design negligently targets minors and results in foreseeable harm. This won’t solve all of the problems posed by the internet, nor should it be used to silence free speech. However, it would finally hold those who designed the product accountable for their actions and their profits.
It’s unjustifiable for a company to make a business model out of a teenager’s five-minute scroll.
