A jury is rewriting the platform accountability playbook in the US

By Dr. Konstantina Bania

A California jury has done something US courts have long refused to do: hold Meta and YouTube legally responsible not for what users post, but for how their platforms are designed to keep children hooked. By finding defective design and negligence and bypassing the traditional safe harbour of Section 230, the verdict marks a structural shift in platform liability; the verdict lines up strikingly with the EU’s Digital Services Act (DSA), which already treats interface and recommender design as a core regulatory concern rather than a neutral backdrop.

Facts of the case

The ruling concerns a Los Angeles case brought by Kaley G.M. against Meta (Instagram) and Google (YouTube). The plaintiff, now in her twenties, alleged that as a child and teenager she developed serious mental‑health harms as a result of compulsive use of Instagram and YouTube, which she said were engineered to exploit vulnerabilities in developing brains.

Key factual elements:

• Claims and defendants: Meta and YouTube were sued on theories of negligent and defective product design and failure to warn, not for failing to remove particular harmful posts. Snap and TikTok were initially part of the broader litigation wave but settled before trial; the LA case went forward only against Meta and YouTube.

• Design features at issue: The plaintiff’s lawyers focused on what they described as a “digital casino” of engagement‑maximising features: infinite scroll, autoplay, algorithmic recommendations, push notifications, streaks, beauty filters and other tools tuned to maximise time on platform. Expert testimony and internal documents were used to argue that the companies knew these features were particularly risky for children, yet continued to deploy and refine them without adequate safeguards or warnings.

• Jury findings and damages: The jury found both Meta and YouTube negligent in the design and operation of their platforms, concluding that these designs were dangerous and that the firms failed to warn users about the risks. Meta was held 70% responsible and YouTube 30%, with around USD 3-6 million awarded in compensatory (and potentially additional punitive) damages.

Both companies have said they will appeal. But regardless of the appellate outcome, the trial has already rewritten the litigation playbook.

Why this case is significant

The verdict has been widely described as social media’s “Big Tobacco moment”, the first time a jury has explicitly labelled mainstream social media products as defective because they are designed to addict and harm children.

Section 230: safe harbour with new holes

For nearly three decades, Section 230 of the US Communications Decency Act has insulated platforms from liability for user‑generated content: they’re treated as distributors, not publishers. Therefore, they are not usually on the hook for what users post. Until recently, plaintiffs who challenged social media harms typically framed claims as “you hosted or amplified harmful content and didn’t take it down,” arguments that almost invariably hit the Section 230 wall.

The Meta/YouTube cases are different in two crucial ways:

• Section 230 does not reach product design: Plaintiffs argued that the harm flowed from design decisions rather than from specific items of third‑party content. Earlier cases like Lemmon v Snap (about a speed‑filter encouraging dangerous driving) had already hinted that design‑based product claims could fall outside Section 230. The LA verdicts take that logic and apply it to the core engagement architecture of social networks. This matters because it opens a lane that many tech companies had hoped to block permanently: liability not as publishers of speech, but as designers and marketers of risky digital products – more akin to car manufacturers or pharmaceutical firms.
• Focusing on design, not (free) speech: The second reason for which the case is significant lies in reframing what is “wrong” with social media. The jury was not asked to adjudicate which posts were harmful, political bias, or censorship. It was asked whether Meta and YouTube knew their design choices could harm children and whether they acted reasonably in light of that knowledge. That shift sidesteps free‑speech arguments. You can defend your right to host controversial speech; it is much harder to defend autoplay and infinite scroll if internal files show you understood these features caused demonstrable dependence and mental‑health harms in minors and chose not to meaningfully adjust or warn.

In effect, the verdict says: the problem is not (only) the speech; it is the system built to monetise attention, regardless of what speech flows through it.

Implications for future cases

The immediate financial impact of these particular verdicts is trivial to trillion‑dollar firms. The legal and strategic consequences are anything but.

A new wave of design‑liability litigation

Commentators note that more than 2,000 similar suits related to social‑media‑induced harms are already in the pipeline in the US. The Meta/YouTube verdict will:

• Validate product-defect theories: Plaintiffs can now point to a jury that accepted the core thesis that mainstream platform design can itself be defective and negligent when targeted at children. TikTok, Snap and smaller platforms may face copy‑cat claims, with plaintiffs probing every engagement‑driving feature (e.g., filters, streaks, rewards, push logic) for evidence of ill‑considered risk.
• Pressure for disclosure: Discovery becomes a strategic weapon. Internal A/B tests, UX research, and risk assessments will be combed for proof that executives balanced engagement metrics against known child‑safety harms and chose engagement.

Even if many suits fail or settle, the litigation threat alone will change the risk calculus in product‑design teams and boards.

Business‑model and design changes

If appeals courts uphold the idea that design choices can trigger liability, platforms may be forced to rethink core engagement mechanics, including:

• Age-segmented design: Features like infinite scroll, autoplay and some recommender default settings could be restricted, slowed or disabled for minors, with stricter age‑assurance tools to justify such segmentation.
• Warning and transparency duties: As with other risky products, courts could demand clearer warnings about the potential for compulsive use and mental‑health impacts, especially for teens and parents.
• Algorithmic safety obligations: Recommendation systems might need to embed safety constraints by design for younger cohorts, limiting exposure to self‑harm, eating‑disorder or extreme‑comparison content, and capping session length.

Such steps are not purely hypothetical. Some have already been floated or partially implemented in response to legislative pressure and public scandals; the verdicts sharply raise the cost of not taking them seriously.

Alignment with the EU’s Digital Services Act and the way forward

Strikingly, what US juries are just beginning to articulate in tort language (i.e., the idea that design choices create legal obligations) is already baked into the EU’s Digital Services Act. The DSA is built on the premise that large platforms are system designers, not passive hosts, and imposes specific duties on that basis, including the obligation to assess “systemic risks”, the duty to evaluate recommender systems and interface design, and then implement proportionate mitigation measures, and the prohibition of employing dark patterns. In other words, the EU has already made design a regulatory object; the Meta/YouTube verdict shows US tort law converging – not on speech, but on architecture. The logic of both the DSA and the LA verdict is simple but powerful: when your code meaningfully contributes to the risk environment, you can’t hide behind a content‑intermediary shield.

For global firms like Meta and YouTube, that means design choices will increasingly be judged against a dual standard: EU regulators’ DSA expectations and US juries’ evolving sense of what constitutes reasonable care in designing attention‑driven systems. Ignoring either is no longer an option.

The image was AI generated.