Using social media safely, and not making it worse for anyone else

Who is writing this, and what they have to gain

The person writing this runs a social platform. TruXSocial is live, people use it, and that is a direct interest in you believing that social platforms can be used well.

Here is what it would look like if that interest were steering this page. The argument would land on users being the problem rather than design, because that is the conclusion that lets an operator off. And somewhere near the end there would be a recommendation to come and use the author's own service instead.

Neither happens here. No service is recommended on this page, and the one place where naming this platform would be the obvious move instead says what it does not do.

A disclosure is not a cure. It is a handle for the reader to hold. If you came here in the middle of something, skip to the next section — the argument can wait.

If something is happening right now

Read this part first if something is already underway. Nothing in it is a judgement about how you got here.

1. Capture it before anything else. The post, the profile, the address bar, the date and time. Deleting removes your evidence, not theirs.
2. Do not reply. A reply confirms the account is live and reachable, and it carries the thing to everyone who follows you.
3. Block after capturing, not before. Blocking can hide the content from you while leaving it visible to everyone else, and you cannot record what you can no longer see.
4. Report it inside the service, and write down the reference number it gives you.
5. Tell one other person today. Isolation is the condition most of this depends on.
6. If there is a threat of physical harm, a child is involved, or money has moved, this has stopped being a platform problem. The sections below say where it goes instead.

There is an exception worth knowing, because the standard advice can make things worse. If the person doing this knows you offline, or lives with you, blocking signals that you noticed, and changing passwords on a shared device is visible. Contact a specialist service for that situation before changing anything.

None of this stops what is happening. It preserves the ability to act on it later, which is what is usually lost in the first ten minutes.

What this post does, and what it does not

This is long, and it is written to be returned to rather than read through.

It goes in this order: the argument, and the strongest objection to it; the five places a report can go; harm aimed at one person; harm aimed at your money; harm aimed at children; harm aimed at everybody; how to post without making the place worse; and the half of the problem this page cannot touch.

There are no jump links inside this page, which is why the parts are listed here in the words their headings use. The last section repeats those words again, so your browser's find can reach any of them.

Every section that tells you to do something ends with a sentence saying what it does not achieve. Advice without its limit is how people end up surprised.

No service is named anywhere in this post. Partly that is a rule this site holds itself to. Mostly it is that naming them dates a piece like this within a year, and lets a reader decide the advice was about somebody else's application.

One more thing about words. Where this page says "you", it means you as somebody posting and deciding. The person on the receiving end is called that, deliberately — because nothing here says a person who was harmed brought it on themselves. Precautions lower what a mistake costs. A mistake is not consent.

Safety advice is written for the person it happens to

Search for advice of this kind and it arrives addressed to a potential victim: lock the account, do not share the address, be careful who you talk to. All reasonable. All aimed at one of several people who decide what actually happens.

Almost all published advice covers only the first, and addresses it to the person it might happen to. This page covers all three, and spends most of its length on the second — the only one a reader can change today without anyone's permission, and the one nobody is ever asked to change.

The second is worth spelling out, because it is invisible from inside. Reposting. Quoting something to condemn it. Screenshotting a thing out of the place it was said into a place it was not. Tagging a location. Replying to bait. Posting other people's faces. Asking whether anyone knows who this is. Passing along a warning that carries the thing it warns about.

Nobody in that list intended anything. That is the point — intention is not an input the system takes.

The weakness in this argument

Everything above is also what an operator would say if they wanted a principled-sounding reason to talk about users rather than about design.

Three concessions, then. The first: whoever runs a service decides how far a pile-on travels, whether strangers can reach you by default, and how quickly a report is looked at. Those are not the reader's decisions, and a theory centred on readers quietly moves them out of frame. The second: a good deal of the worst harm is not ordinary at all. A person grooming a child is not behaving normally, and advice about ordinary behaviour does nothing whatsoever to a fraud operation with a staff rota. The third: this page is addressed to people with reading time and a measure of safety, and the people most exposed to all of it usually have less of both.

So the claim retreats to the defensible version: the second of those three things is large, it is under-discussed because nobody profits from discussing it, and it is the part a reader can act on immediately. That is smaller than where this section started.

A theory that flatters the person who wrote it should be read at a discount. This one does. Read it at one.

The five places a report can go

These are described once, here, and referred to by number afterwards. Most people use the first and stop, and the first is the weakest of the five for exactly the harms that matter most.

One: inside the service

The report control on the post or the profile. Good at getting a thing removed. Poor at identifying anyone, poor at speed, and impossible to argue with. Keep the reference number it gives you; it is what makes the later steps possible.

What it does not do: it usually does not create a record anybody outside the company can use. Material involving children is the significant exception — providers in some countries are required to pass those reports to a national body, and the overwhelming majority of what that body receives arrives from providers rather than from the public. For that one category, reporting inside the service is the strongest of the five rather than the weakest.

Two: the service's specialist route

Larger services run separate routes for child safety, for intimate images, and for law enforcement, and those routes are staffed differently from the general queue. Find one by looking for the service's safety centre or help centre, or by searching the service's name along with the word for the harm.

What it does not do: it does not exist everywhere, and on a small service it is the same person reading both queues.

Three: a national reporting body

Most countries have one for online crime, one for child protection, and one for fraud, and they are frequently three different bodies. This page does not print a list of them, because such a list rots quietly and a reader cannot tell when it has.

The durable method instead: search for the name of your country together with the words "report" and the harm — "child protection", "cyber crime", "online fraud" — and prefer the result on a government domain. Your national police service's website will name the right body even when it is not the police themselves.

For material involving children specifically, there is a shortcut that removes the guessing. A network of national hotlines covers much of the world, and its site lists member countries with a link to each one's reporting route. Finding yours there is faster and more reliable than searching, and it means the advice in this post works wherever you are reading it rather than only in the country a given service happens to sit in.

What it does not do: the body that takes the report is frequently not the body that can act, and being passed along once is normal rather than a sign of failure.

Four: the police

Genuinely appropriate when there is a threat to life, a child involved, money that has moved, stalking, or extortion. Bring what you captured, the reference number from step one, and a plain sequence of what happened and when.

What it does not do: jurisdiction is a real obstacle when the other person is in another country, and it is frequently the reason a case stops.

Five: the services that work across every platform

These are worth knowing about because they operate across services rather than inside one, and because they are the least widely known thing on this page.

Both of the ones below work the same way, and the way they work is much less alarming than it sounds. Neither asks you to send anybody the image. Your device calculates a fingerprint of it — a hash — and only that fingerprint is shared, so participating services can recognise the image without ever holding a copy.

If the person in the image was under eighteen when it was taken: Take It Down, which is free. In its own words it turns the image into "a unique digital fingerprint, called a hash value", and — the sentence that matters most to a frightened person — "Your image or video remains on your device and is not uploaded."

If they were eighteen or over at the time, and are now: StopNCII, also free, available in a great many languages and not restricted to one country. It states the same guarantee in the same terms: the hash "is the only file which is shared", and it "does not download the images from your device."

What neither of them does: they reach only services that have agreed to take part, and only content that is publicly reachable. As Take It Down's own page puts it, platforms "may have limited capabilities to remove content that has already been posted in the past." Private messages and encrypted services are outside their reach entirely.

Harm aimed at one person

Threats

What separates a threat from an argument is that it describes something happening to a specific person, rather than describing an opinion about them.

Capture first, including the account and the timestamp. Rung one to get it removed, rung four if it names a place, a time, or anything that suggests the person knows where you are. A threat that contains a detail about your actual life is a different category from one that does not.

What that does not do: removal is not investigation, and a removed post is a destroyed copy of your evidence unless you captured it first.

Coordinated abuse and pile-ons

The distinguishing feature is volume rather than content. One person being rude is one person being rude; several hundred arriving within an hour is a different thing, and it frequently began somewhere the person on the receiving end cannot see.

Capture the pattern, not only the worst message — the timing and the repetition are the evidence. Rung one for individual messages, rung two if the service has a route for organised campaigns.

What that does not do: reporting the messages one at a time is close to useless at this volume, and being told to block them individually is advice that has not understood the problem.

If someone has published your address, workplace or school

Speed matters more here than anywhere else, because the risk is that somebody acts on it.

Capture, then rung one immediately, then rung four if it is your home. Tell the place that was named — the workplace or the school — before somebody else does, because they respond very differently to a warning than to a surprise.

What that does not do: removal from one place does not remove copies, and the realistic goal is limiting spread rather than erasure.

Someone is pretending to be you

Impersonation accounts are usually built from material you posted publicly, which is why they are convincing.

Rung one, using the service's impersonation route rather than the general one, since it usually asks for identification and moves faster. Tell the people likely to be approached — that is what actually stops the damage, and it works before any report is processed.

What that does not do: the account being removed does not tell you who made it, and a second one frequently appears.

Intimate images shared without consent

This is the one where the cross-platform route in rung five matters more than the platform-specific one, and where doing rung five first is usually correct.

What that does not do: it does not reach private messages, encrypted services, or any platform that has not joined the scheme.

Stalking, and when the person is known to you

Everything above assumes the other person is a stranger. When they are not, some of it is actively wrong.

That is not a cautious guess. Guidance published for people being monitored states it directly: removing monitoring software, or "making significant changes may be detected by the abuser and could increase the abuse and harassment." Its advice is to make a plan with a specialist service first, from a different device, and to act only once that is safe.

Keep a log with dates rather than a folder of screenshots — a pattern over time is what is useful to the police, and single incidents individually look minor.

What that does not do: none of this addresses the underlying situation, and it is not a substitute for a service that handles it.

Harm aimed at your money

The shapes fraud takes

Described by mechanism, because the branding changes constantly and the mechanism does not.

• The long relationship that eventually needs money, built patiently over weeks.
• The investment with a convincing dashboard, where small withdrawals succeed early.
• The job offer with an upfront cost, or one that asks you to receive and forward money.
• The marketplace deal that moves off the platform to a payment method with no protection.
• The appeal after a disaster, which arrives while everyone is generous and distracted.
• The recovery service that contacts you after you have been defrauded — the second fraud, run on the same person, and frequently by the same people.
• The support call about a problem with your account that you had not noticed.
• The message from a friend whose account was taken.

The two tests that catch most of it

Two questions, and they are about the shape of the conversation rather than its content.

Is there urgency? Manufactured time pressure exists to stop you consulting anyone. Is it moving channels? A request to continue somewhere else — a messaging app, a different service, a phone call — is the moment the protections of the first place stop applying.

What those do not do: they catch the common shapes and miss the patient ones. Somebody willing to spend four months is not going to fail an urgency test.

If money has already moved

Order of operations, and the order is the whole point:

1. The bank or card issuer, first. Speed is the only variable that meaningfully changes the outcome, and it degrades by the hour.
2. Rung three or four, for the report.
3. Rung one, for the account.

What that does not do: recovery is uncommon once money has left, and anybody who contacts you promising to recover it is the next fraud.

The part addressed to people who are not being scammed

Replying "this is a scam" underneath the post shows it to everybody who follows you. Screenshotting the bait to warn people spreads the bait, in a form that has been stripped of its context and made more shareable.

Report it and say nothing. The warning that reaches people is the one you send to the person you think is about to fall for it, privately.

Harm aimed at children

The frame this section does not use

There is a version of this subject built entirely out of fear, and it does not work — not because it is unkind, but because of what it produces.

Fear produces surveillance. Surveillance produces concealment. Concealment produces exactly the isolation that every harm in this section depends on. A young person who believes an adult will overreact handles it alone for days, and the days are what turn a bad situation into a much worse one.

So this section describes mechanisms rather than scenarios, and the most useful thing on it is not a setting.

Grooming, described by its mechanics

It has a shape, and the shape is recognisable in a way that individual messages are not.

• Attention, and a lot of it, at a point when the young person is short of it.
• A secret, early and small, that they now share with this person.
• Distance from other people, introduced gently — those people would not understand.
• A move to a private channel, or one where messages disappear.
• A first small compromise, which is used to prevent them telling anyone about the second.
• Escalation, once telling anyone has been made to feel impossible.

A child who has been told this shape can recognise it happening. A child who has only been told to be careful has been given nothing to recognise.

What that does not do: recognising it does not make it easy to act on, and the reason it continues is rarely that the young person failed to notice.

Sextortion of teenagers

The pattern is a relationship that becomes intimate quickly and then becomes blackmail, and it moves faster than adults expect — sometimes within a single evening.

The guidance from the American centre for missing and exploited children begins by contradicting exactly what fear tells a frightened person to do: "Cooperating or paying rarely stops the blackmail." Its instructions are to report the account through the platform's own safety feature, and to "BLOCK the suspect but DO NOT DELETE your profile or messages because that can be helpful in stopping the blackmailer." Their own reporting route is the CyberTipline, which is the American one; anywhere else, rung three names the equivalent for your country. Rung five is how the images themselves get taken down, and it works regardless of where you are.

The sentence that belongs with all of it, in that centre's own words: "the blackmailer is to blame, not you."

What that does not do: it does not undo the fear, and it does not work at all while it stays a secret. This stops when it stops being secret.

If you find material involving a child

This is the one place on this page where handling something yourself makes it worse rather than better, including legally.

Accounts belonging to children too young to have one

A minor on a service with an adult minimum is the normal case rather than the exception. Age rules are typed in, and a determined child types a different number.

This is the second and last time this platform is mentioned, and it is mentioned as a limit: TruXSocial's own minimum age is not settled, and rather than print a plausible number, its page says so and leaves the question visibly open.

What that does not do: an unsettled question is not a solved one, and saying so publicly does not protect anybody by itself.

Posting other people's children, including your own

The problem is not any single photograph. It is assembly.

A face, a school uniform, a location tag, a routine mentioned twice, a name, and a birthday — published affectionately over four years — is a complete profile that nobody else had to do any work to build.

Specific rather than "be careful": crop out the uniform and the street sign, post the holiday after coming home rather than during, leave the full name and the birth date off, and ask the other parent before posting a photograph of their child.

What that does not do: it does not retrieve anything already posted, and a child who later objects has no route to undo it.

Harm aimed at everybody

Recruitment, and the jobs that end in a locked building

Job advertisements on social platforms are a recruitment channel for fraud operations, and some of the people running the frauds described earlier were themselves recruited this way and cannot leave.

The markers are consistent: unusually high pay for unclear work, a job abroad with travel arranged for you, an interview conducted entirely inside a messaging application, and pressure to decide quickly. Anyone being recruited across a border with the arrangements handled for them should tell somebody who is not involved exactly where they are going.

What that does not do: this is a description, not an intervention, and it is of little use to somebody with no other options.

Illegal trade in the open

Places where strangers can reach each other are used to sell illegal drugs, including counterfeit tablets pressed to look like ordinary medicine — which is the part that kills people who believed they knew what they were taking.

This page deliberately does not describe how those sales are advertised, the coded terms used, or how the accounts avoid review. Publishing that informs the wrong readers, and the transparency page sets out the reasoning in full.

Report it at rung one, and rung three if it involves a young person you know.

What that does not do: accounts of this kind are replaced quickly, and reporting one is not the same as stopping the trade.

Someone posting about hurting themselves

This is the one place on this page where reporting is not the first move.

Respond first, privately, and plainly — say you saw it, and ask directly. Then stay with it long enough to get them to somebody who is qualified, which usually means a crisis line in their own country. Do not repost it, do not quote it, and do not discuss the method.

What that does not do: you are not a clinician, and the goal is not to resolve it yourself. It is to be the person who did not scroll past.

The warning that carries the payload

During anything fast-moving, the misinformation that travels furthest is frequently somebody warning about misinformation, with the original attached.

The reader-side job is not to become a fact-checker. It is to stop being a repeater. The correction never travels as far as the claim, and it never has.

How to post without making the place worse

This is the half with no settings page, and it is where most of a reader's actual influence is.

What you publish about other people

A screenshot of a private conversation involves a second person who agreed to none of it. Footage of somebody being hurt or humiliated still belongs to them, and their family will see it. A person in the middle of a crisis is not content.

None of that is enforceable on you, which is exactly why it is worth writing down.

Amplification is an action

Passing something on is a decision, not a neutral act.

The question before resharing is not "is this interesting?" but "what happens to the person in it if this is wrong?" That has a real answer, and it takes about four seconds.

The pile-on you are in

A reply that is one of five is criticism. The identical reply as one of five thousand is something else entirely, and the person sending it cannot see which one they are sending. The harm is produced by the total, and no individual contributor ever feels responsible for the total.

If somebody is already buried, adding to it changes nothing except the weight. Nobody has ever been argued into a better position by the ten-thousandth message.

The wrong name

During an incident, names circulate before facts do, and ordinary people are misidentified. Naming somebody you have not verified is not participation in justice — it is the production of a new victim, and they have no route to undo it.

What you publish about yourself

Deliberately the shortest part of this section, which is the argument of this page expressed as a layout.

• Keep a different password on the mail account than anywhere else, because that account can reset all the others.
• Use two-step verification wherever it exists. As the British national cyber security body puts it: "Even if a criminal knows your password, they won't be able to access any of your accounts that are protected using 2SV." Prefer an authenticator application over codes by text where both are offered.
• Treat privacy settings as a decision about audience, not about permanence. Anything visible to somebody can be captured by them.
• Apply the test that body suggests before posting: consider "what your followers and friends need to know, and what detail is unnecessary (but could be useful for criminals)".

What that does not do: none of it defeats a determined and resourced person, and it is not a guarantee of anything — those are targets at best, and this site does not print promises nobody can keep.

The half this post cannot do

The third of the three things named at the start belongs to whoever runs the service, and nothing on this page touches it.

Which posts get carried furthest. Whether strangers can reach you without being invited. How quickly a report is read, and by how many people. Whether rules are published before they change. Whether a removal comes with a reason and an appeal. Whether the thing is built to be hard to put down.

Those are decisions made by operators, and a post by an operator about what users should do is a post about the easy half. That is a fair description of this page.

There is one demand worth making of any service, and it should be made of this one too. When you report something, you should be told that the report arrived, what happens next, and what was done about it. That is not an unusual ask: the standard Samaritans publishes for the technology industry sets those three as the minimum — acknowledgement, "Information about what happens next", and "Action taken". Most services meet none of them, and a report that disappears without a word teaches people to stop reporting. Nothing on this page should be read as a claim that the author's own platform clears that bar.

What can be checked instead is whether the operator wrote down what they refuse to do, before there was anything to lose by writing it. That list, including the promises refused as unkeepable, is on the principles page, and it is the part with a date on it.

What to remember

There are no links inside this page. Each line below repeats the words of the heading it came from, so your browser's find will reach it.

• If something is happening right now — capture before you block, do not reply, tell one person today.
• The five places a report can go — inside the service, its specialist route, a national body, the police, and the services that work across every platform.
• Threats — a detail about your actual life makes it a different category.
• If someone has published your address, workplace or school — tell the place that was named before somebody else does.
• Stalking, and when the person is known to you — the standard advice can escalate it.
• The two tests that catch most of it — urgency, and moving channels.
• If money has already moved — the bank first, because only speed changes the outcome.
• The frame this section does not use — fear produces surveillance, surveillance produces concealment, concealment produces isolation.
• Sextortion of teenagers — it stops when it stops being secret.
• If you find material involving a child — do not download, forward, investigate or confront.
• Posting other people's children — the individual posts are harmless; the collection is not.
• Amplification is an action — quoting something to condemn it still shows it to everyone who follows you.
• The wrong name — naming somebody unverified produces a new victim.
• The half this post cannot do — the operator's half, which this page does not settle.

Safety is not a setting. It is what a few thousand ordinary decisions add up to, most of them made by people who are not in any danger themselves.

Sources

• The American centre for missing and exploited children, on sextortion
• Take It Down, for images taken of someone under eighteen
• StopNCII, for images taken at eighteen or over
• The network of national reporting hotlines, for finding the route in your own country
• The CyberTipline, the American reporting route
• The British national cyber security body, on using social media safely
• Welsh government guidance on victim-blaming language
• Guidance for people being monitored, from the coalition against stalkerware
• Samaritans, on what reporting should tell the person reporting