Should Social Media Platforms Be Responsible for Their Content?

A platform that decides what billions of people see is not a neutral pipe. Telephone lines carry every call indifferently; a newsfeed ranks, recommends and demotes, and it earns more when sensational content spreads. The real question is therefore not whether bad content should exist, but whether the company that amplifies and monetises it owes a duty for what it engineers. I argue that it does, provided that responsibility is bounded by proportionality and due process rather than turned into open-ended liability for every user post.

The first reason is that amplification is an editorial act. Once a platform's algorithm chooses who is heard loudly and whose speech is buried, it is making the kind of decision a publisher makes, and the 'we merely host it' defence understates what it actually does. The European Union recognises this: its Digital Services Act designates the largest services as 'very large online platforms' and treats them as systemic-risk actors subject to independent audits. With algorithmic power comes algorithmic responsibility; freedom of speech is not the same thing as freedom of amplification. Therefore platforms should be responsible for the content they boost.

The second reason is that the harms are systemic, lethal, and otherwise unaddressable. These are not abstract upsets. The 2019 Christchurch attack was livestreamed and Facebook removed roughly 1.5 million copies in the first twenty-four hours; the United Nations fact-finding mission found that Facebook had played a 'determining role' in the violence against Myanmar's Rohingya; and an English coroner concluded that online content contributed to the death of fourteen-year-old Molly Russell. Falsehood also travels faster than truth: an MIT study published in Science found false stories were about seventy per cent more likely to be retweeted. Individual takedowns cannot answer a problem that is structural, so only a platform-level duty of care can. Therefore platforms must be responsible.

The strongest objections must be met, not dodged. Platforms are private firms, not states; in the United States, Section 230 immunises them precisely so they are not treated as the publishers of user speech, and deputising them as censors gives unelected companies power over truth. Worse, 'harm' and 'misinformation' are vague and weaponisable: blunt automated filters mislabel satire and dissent, and governments can lean on platforms to silence critics under a safety banner. Heavy compliance can also entrench incumbents, since only giants can afford risk audits. These are real risks of responsibility done badly.

They are answered by calibration and due process, not by abdication. Responsibility should target behaviour and amplification, not belief: provenance labels, throttling of inauthentic coordination, and demotion of virally boosted falsehood, while protecting the underlying right to be wrong. Duties should be tiered by size, as the DSA does, so start-ups are not crushed. And moderation must carry administrative-law virtues, notice, reasons, appeal and transparency. The Facebook Oversight Board upheld Donald Trump's 2021 suspension yet condemned the indefinite penalty as 'arbitrary' and 'standardless', which is exactly the unaccountable private power that formal responsibility cures. The European Court of Human Rights confirmed in Delfi v Estonia (2015) that holding a platform liable for extreme user comments can be compatible with free expression when proper safeguards exist.

Comparative practice maps onto this line. The United Kingdom's Online Safety Act 2023 imposes a statutory duty of care enforced by Ofcom, with fines up to ten per cent of global turnover; the EU's risk-tiered DSA pursues the same end more granularly; Canada attempted a similar regime in Bill C-63 before it lapsed on prorogation in 2025; and the American Section 230 model sits at the permissive extreme. Article 10 of the European Convention frames the underlying right as qualified, exercised 'with duties and responsibilities' and limitable where 'necessary in a democratic society'. That is the correct frame here too.

Platforms should therefore be responsible for their content. Drafted broadly, such responsibility chills lawful speech and entrenches monopolies; drafted narrowly around amplification, with proportionality and due process, it converts arbitrary private governance into accountable governance. The honest choice is not between perfect moderation and none, but between accountable imperfection and reckless abdication, and abdication is the costlier mistake.