What the other person sees when you filter messages

It is one of the most common worries people have before they start filtering a difficult ex, relative, or contact: “What will they see?” For many people the whole point is to lower the temperature, not to provoke a reaction by appearing to block or ignore someone. This guide sets out exactly what reaches the other person, and just as importantly, what does not.

The short answer

They get one short notice, once. The first time someone messages your relay address or number, they receive a single line telling them messages are processed by AI and stored, with a link to the privacy page. After that, the only sign is a short “Sent via FenceChat using AI” line on messages you send through FenceChat. Every later message goes through with no notice at all, and no indication that a particular message was changed.

Being straight about this matters. A service that collects and stores what people write, and tells them nothing ever, sits uneasily with Australian privacy law. One neutral notice on first contact is both the honest position and the more defensible one.

What it looks like from their side

Picture a difficult ex sending you a message. They open their normal texting app or email, type their message to the contact they have for you, and hit send. On their screen it shows as sent, then delivered, exactly as it always has. While your account is active there is no bounce, no error, and no “this person is unavailable”.

The first time only, they also get a short notice. It is deliberately plain: it names the service, says messages are processed by AI and stored, and links to the privacy page. It does not say who you are, does not say anything was removed, and does not invite a reply.

What they do not see is what happens next. Their message arrives at your relay, where it is screened before anything reaches you. You receive the parts that matter and are shielded from the hostility. None of that is visible to them, because none of it happens on their device. They are never shown a redacted version, never told which words were removed, and never told whether a given message was altered at all. The screening is done by an AI message filter.

Why one-sided is the part that matters

The thing that makes this possible is that the filtering is one-sided. Many tools for difficult relationships are two-sided: both people have to install the same app and agree to communicate through it. The moment you ask someone to download an app to talk to you, you have told them something has changed, and a hostile person may well refuse or use it as ammunition. This is the main difference from a typical co-parenting app.

A one-sided relay removes that entirely. Because the tool lives only with you and works over ordinary email and SMS the other person already uses, there is nothing for them to install, nothing to agree to, and no cooperation to negotiate. That is the part that actually protects you: it works whether or not they are willing to be reasonable. A one-line notice does not change that. Being told a service exists is very different from being asked to join one.

Why this is different from blocking

Blocking and filtering are often confused, but they feel very different to the person on the other end. Blocking tends to leave clues: calls may go straight to voicemail, messages may never show as delivered, and a determined person who is paying attention can often work out they have been blocked. For some people that realisation triggers exactly the escalation they were trying to avoid, including switching to new numbers or other channels.

Filtering through a relay avoids this because the message is still received. Nothing is bounced or refused. The contact still works from their point of view, so there is no sudden wall for them to notice and react to. You can read more in our guide on filtering versus blocking.

When the notice can be switched off

The notice is on by default, because for most people it is harmless and being open about the service is the honest approach. But not every situation is ordinary. If you are hiding from someone, an email telling them a service is in use can be the thing that provokes them.

So it can be switched off, per account, at any time. FenceChat asks once during setup and the choice sits in Settings after that. The judgement belongs with you, because you are the only person who knows what that contact is capable of.

Beyond the notice, your own replies can still give things away. Two practical points:

• Keep replying when a reply is genuinely needed. If you used to answer logistics questions and suddenly go completely silent, a person may notice the change in pattern. Brief, calm replies to the things that matter keep everything looking normal.
• Give out the relay contact, not a sudden “new number” announcement. The smoothest approach is to have them use the relay address or number from the start, so there is no visible switch.

None of this is about deceiving anyone. For most people the goal is simply to stop absorbing abuse without lighting a fuse. A single plain notice, then a quiet filter, is the calmest way to do that.

How FenceChat does it

FenceChat is built entirely around the one-sided model. You give the other person an ordinary email address or phone number to reach you on. They message it like any other contact, installing nothing and signing up to nothing. FenceChat filters what comes in, passes you what matters, and keeps a complete archive, all on your side. Our guide on how FenceChat works walks through set-up.

They get one plain notice the first time they write to you, and after that only a short “Sent via FenceChat using AI” line on messages you send through FenceChat. Because the filtering happens after their message arrives, no individual message is ever flagged to them as changed. You get to lower the temperature, protect your peace, and keep a record, on your own terms.