Leaked Facebook rules, UK law and Valve's lawyer: moderation's uneven decade

The documents, more than 100 internal training manuals, spreadsheets and flowcharts, were passed to the Guardian and published on the night of 21 May. Ars Technica reported the following day. They set out how the company handled hate speech, violence, self-harm and terrorist propaganda across a user base then approaching two billion people worldwide.

Under the leaked rules, a post reading "To snap a bitch's neck, make sure to apply all your pressure to the middle of the throat" was allowed to stay up. So were comments such as "kick a person with red hair" or "let's beat up fat kids." A post carrying the message "Someone shoot Trump" was banned, with moderators advised to remove it. Ars Technica noted that threatening a president is a criminal act in the US.

The guidelines explained the logic:

We aim to allow as much speech as possible but draw the line at content that could credibly cause real world harm. People commonly express disdain or disagreement by threatening or calling for violence in generally facetious and unserious ways.

Facebook's then global policy management boss, Monika Bickert, told the Guardian that policing a "diverse global community" was impossible. The line between satire and inappropriate content was "sometimes very grey," she said. The company said at the time it was hiring an extra 3,000 content moderators to crack down on hate speech and child exploitation. It refused to say whether the jobs would be in-house or outsourced.

Nearly six and a half years later, the UK passed its own answer. The Online Safety Bill received royal assent on 26 October and became the Online Safety Act, The Verge reported. It obliges tech firms to address underage access to online pornography, anonymous trolling, scam ads, nonconsensual intimate deepfakes, and the spread of child sexual abuse material and terrorism content.

Enforcement is phased. Ofcom, the UK telecoms regulator, planned to publish codes of practice in three stages, starting with illegal content such as terrorism and child sexual abuse material, with a consultation due on 9 November. Penalties reach up to £18 million, around $22 million, or 10 percent of global annual turnover, whichever is higher, with bosses potentially facing prison. Messaging apps including WhatsApp and Signal objected to a clause letting Ofcom ask companies to identify child sexual abuse content "whether communicated publicly or privately," saying it fatally undermined end-to-end encryption. Some providers suggested they would rather leave the UK than comply.

Valve's internal argument

Gaming platforms have run the same argument on different terms. PC Gamer reported on 2 June 2026 on a Bloomberg account, attributed to an anonymous former Valve employee, of a dispute between co-founder Gabe Newell and general counsel Karl Quackenbush. Quackenbush was advocating a more hands-on approach to content moderation on Steam as the company considered embracing outright pornographic content.

Newell's reported reply was blunt: "What the fuck do I pay you for if that's your opinion?" PC Gamer notes that no further details of the exchange are available, but that the outcome is visible: Steam did open up to NSFW content, and its moderation approach remains laissez-faire, apart from the occasional clash with payment processors. Valve delisted adult content from the platform the previous year following pressure from those processors.

The three episodes point at the same unresolved question. Facebook spent years arguing it was a platform rather than a publisher, a position that shielded it from libel exposure and leaned on the EU E-commerce Directive, whose Article 15 says hosting services are not obliged to monitor what they store. The UK answered with statutory duties and fines. Valve answered with a shrug. What none of them settled is where the line sits, and who is accountable when the grey area turns out to be someone's real-world harm.