In Singapore, a woman knew her harasser. In China, 20 women never met theirs

Alanna (pseudonym) is 31. After a fallout with a former colleague, the latter began posting her photographs online without her consent, accusing her of being a narcissist and having mental issues, and signing up her personal and work email addresses on websites such as Pornhub.

She took screenshots. She gathered statements from friends that he had contacted, one of whom was able to identify him.

She made a police report and applied for a protection order under the Protection from Harassment Act. These were both ultimately dismissed as she was unable to prove that the ex-colleague was behind the harassing accounts.

“It was frustrating that I couldn’t prove it, even though I knew it was him,” she told researchers from SG Her Empowerment in an interview for a study that the Institute of Policy Studies released last October. “I feel a strong sense of injustice that he was able to do all these things… hiding behind his laptop.”

Online, harm arrives long before a name does, and often no name arrives at all.

Singapore: find the person

Singapore’s answer to helping people seek recourse for online harms began operations on 29 June. The Online Safety Commission is one of a few agencies anywhere dedicated to victims of online harm. It can order platforms to remove content or restrict accounts, and filing a report is free, requiring neither a lawyer nor a court order. In its first phase, the OSC focuses on five online harms: intimate image abuse, image-based child abuse, doxxing, online harassment (including online sexual harassment) and online stalking.

The Act behind it, the Online Safety (Relief and Accountability) Act, specifies statutory tort provisions allowing victims to seek remedies from the court. A victim who intends to sue can ask the Commission to find out from the platform who was behind the account.

Minister for Digital Development and Information Josephine Teo put it plainly in Parliament last November. Victims, she said, “are often unable to confirm who was responsible for the harm. They continue to be in fear and cannot find closure.” The Act “provides victims with a means to identify the person who caused the harm”.

For Alanna, it would have opened a door that was shut: an order to the platform to say, or to find out, who was behind the accounts.

China: slow the crowd

In March, China’s Supreme People’s Court published as a model judgment the case of a secondary school teacher in Jiangsu, named only as Zhu, who fell out with a relative. The relative complained to an acquaintance, who proposed a way to get back: buy Zhu’s personal data and smear him online. The acquaintance paid 13,150 RMB for 1,442 personal records: hotel registrations, air tickets, and rail bookings.

Only 299 concerned Zhu and one other target. The remaining 1,143 belonged to strangers, bought in bulk because they happened to overlap with Zhu in time and place.

From that haystack, the pair picked out women who had stayed at the same hotels as Zhu in the same periods, and wrote them into the post as proof of an affair. More than 20 innocent women were implicated, among them a pupil at Zhu’s own school in her final year, with the gaokao weeks away. Reads, shares and replies soon passed two million. The two men were convicted of buying personal information and sentenced to 11 months and a year, respectively.

Consider those 20 women. Nobody was truly aiming at them. Their records were bought to make one man’s lie look true. The court named and punished the men, but that returns nothing to a schoolgirl whose hotel registration was published as evidence of a non-existent affair.

China’s draft Anti-Cyberviolence Law (反网络暴力法), whose public consultation closed on 28 August, starts from the other end. It builds on the 2023 judicial guidance from the courts, prosecutors and police on punishing online violence and on the rules that have bound platforms since August 2024, which already require much of what follows.

Under Article 13, providers must build detection systems and warning models. On finding a risk of online violence, they must stop pushing the material through recommendation algorithms, limit its traffic, label it, and, if it is climbing, report it. They also need to undertake real-name identity verification of the irregular accounts.

Three men make a tiger (三人成虎)

Pang Gong, about to leave for a rival state, puts a question to his king. If one man says there is a tiger in the marketplace, do you believe him?” “No.” “Two men?” “No.” Three? “I believe it,” says the king. “There is plainly no tiger,” says Pang Gong, “yet three men speaking make one. Where I am going is further away than the marketplace, and those who will speak against me number more than three.”

When Pang Gong returns, he is never granted an audience again. The king believes the rumours against Pang Gong because enough people say them, yet no one in particular is responsible.

Han Fei filed the story among his teachings for kings: check what you hear against many sources. His larger remedy was not to chase the many. It was to match name to reality (xingming 形名): authority must be able to hold a specific person to a specific claim. Duty is assigned in advance to someone who can be named, rather than a culprit hunted down afterwards. Online, the one who can always be found is the platform. A net, Han Fei wrote, is drawn by its rope, not mesh by mesh.

Varied approaches, similar aims

Two men answered for buying data. Nobody answered for what happened to the girl, because by the time anyone could, it had already happened. Alanna needed a name. The schoolgirl needed the crowd stopped. Victims need both.

In Singapore, when a crowd turns on someone, a platform’s duty to act mostly begins with a complaint. In China’s draft, it begins when the platform’s own warning models flag a risk, without waiting for anyone to complain.

The platform is the rope in Han Fei’s net. It sees a crowd forming before anyone else does, and it decides what the crowd sees next.

Why has it not already acted? To a platform, a crowd looks like success. In 2025, Meta earned 98% of its revenue from advertising. Your attention is what it sells. A crowd hounding a stranger is also hard to tell from one with a fair grievance, and slowing the wrong one invites charges of censorship.

So most wait to be told. The company said in January 2025 that it would act on “less severe” violations only once someone reported them. Once told, platforms still take days. When IMDA tested Singapore’s designated platforms in 2025, they took two to five days, on average, to act on user reports.

Singapore’s Code of Practice and China’s 2024 rules both require platforms to offer protective settings, such as limiting who can message a user or comment on their posts. China’s rules also require a platform that spots a user at risk to alert them and tell them how to protect themselves. In both countries, protection exists. The hard part is reaching them before the crowd does.

A brake on crowds must also be answerable to users. China’s draft would require platforms to tell users why their accounts were restricted and let them appeal. The European Union already requires reasons even when a post is merely demoted, and requires platforms to publish how many of their decisions were reversed on appeal.

Beyond the platforms

Communities can reach where platforms cannot. Schools in Singapore already teach cyber wellness, guided by principles that include “positive peer influence”. In practice, it means standing by a classmate when a crowd turns on her. Groups such as SG Her Empowerment, whose researchers heard Alanna’s story, offer victims help before they have a name to give.

The rest falls to us, because the crowd is made of individuals. Those who join an online crowd often believe, as the Jiangsu post invited them to, that they are upholding morality by punishing a wrongdoer. Before repeating an accusation against a stranger, ask what you have checked it against. If the answer is only how many others are saying it, don’t add your voice. If you see a crowd turning on someone, report it. You may see it before she does.

Whoever rallies a crowd may also come to answer for it. Singapore’s Act lists “online instigation of disproportionate harm” for a later phase, and China’s draft holds organisers and inciters liable.

Three men make a tiger. Singapore has built a law to find the three. China has drafted one to keep the third from being heard.

Related: Too fast to censor: How online rage is turning on China’s elites | Why hyper-efficient Singapore is Asia’s most AI-anxious society