Contraventions of the newly enacted UK Online Safety Act are now under judicial scrutiny in four separate cases, each raising distinct questions about platform responsibility. The Act, which came into force earlier this year, imposes a legal 'duty of care' on social media companies to protect users from illegal and harmful content. Non-compliance carries penalties of up to 10% of global turnover or £18 million, whichever is greater.
The first case concerns a teenager who died by suicide after viewing pro-anorexia and self-harm content on an algorithmic feed. Her parents argue that the platform's recommendation system constituted a foreseeable harm, a claim the company contests by citing its terms of service. The crux here is whether 'user-to-user' services are liable for algorithmic amplification of user-generated content that falls short of illegality.
In the second case, a terrorism survivor is suing a video-sharing platform for hosting a livestream of the perpetrator's attack. The platform removed the video within 30 minutes, but the claimant argues that even transient availability causes psychological trauma. The Act's provisions on 'priority illegal content' shift the burden to platforms to demonstrate proactive removal, not merely reactive moderation.
The third case involves a data breach. A social network suffered a leak of users' location data, which was then used by a stalker to track and assault a victim. The Act's expanded definitions of 'user safety' now cover data protection if reasonable security measures were not implemented. The company claims the breach resulted from a third-party API vulnerability, not a direct failure of its own systems.
The fourth and most complex case examines the new 'illegal content' categories: the platform knowingly hosted fraudulent advertisements that used deepfakes of a celebrity to promote a scam investment scheme. The celebrity is suing for failure to verify advertisers' identities, a requirement under the Act. The platform argues that AI-generated content is too novel for current verification systems to reliably detect.
These cases are not simply legal arguments; they represent a high-stakes experiment in internet governance. The UK is the first major economy to impose a systemic duty of care on platforms. If the courts interpret the Act broadly, platforms may be forced to redesign their core infrastructures, including algorithm transparency, content moderation pipelines, and user verification layers. A narrow interpretation could render the legislation symbolic.
The outcomes will also influence the EU's Digital Services Act enforcement and potential US federal legislation. The international tech community is watching closely. What happens in these London courtrooms could reshape the global internet's architecture.
From a technological perspective, the core challenge is the sheer scale of content. Platforms process petabytes per second. Even with advanced AI moderation, false positives and false negatives are inevitable. The Act does not require perfection, but a proportionate use of 'best available technology'. This standard is likely to evolve as detection tools improve.
There is also a tension between user privacy and safety. End-to-end encryption, hailed as a privacy safeguard, complicates content moderation. The Act gives Ofcom the power to require platforms to use 'accredited technology' to scan for child sexual abuse material, effectively breaking encryption for that purpose. A challenge to this provisions is expected.
These four cases will set critical precedents. For the families involved, justice is paramount. For society, the question is whether we can design a digital public square that minimises real-world harm without sacrificing free expression. The judges must balance these opposing forces with the precision required by law.
I have seen court documents and expert testimonies. The technology does not yet exist to perfectly filter every harmful post. But the Act does not require perfection. It requires a demonstrable effort, a shift from 'we didn't know' to 'we should have known'. That shift is what these cases are about.
The hearings continue next week. A verdict may take months. Regardless, the implications are clear: the era of platform impunity is ending, but the transition is fraught with legal and technical complications that only time and careful jurisprudence can resolve.
