AI Message Filter

AI message filter:
read the message.
Not the abuse.

Every incoming message is processed by AI before you see it, taking out manipulation and hostility and flagging any threat, while preserving anything genuinely useful. The original is always archived.

What filtering looks like

The same message: what was sent, and what reaches you.

Original message, archived

"You’re completely incompetent and you’ll regret this in court. Pick up is at 4pm Thursday. Don’t even think about being late again or I’ll have my lawyer all over you."

Filtered, delivered to you

"Pick up is at 4pm Thursday."

Factual content is preserved, with hostility removed.

The FenceChat filter screens for

Filtering goes far beyond keyword matching. Our AI even understands tone, context, and intent.

Threats and intimidation

Explicit threats, implied consequences, references to legal action used as pressure rather than genuine communication.

Emotional manipulation

Guilt-tripping, gaslighting language, accusations designed to provoke a reaction rather than communicate information.

Verbal abuse and insults

Name-calling, derogatory language, personal attacks, and statements intended to demean or degrade.

Harassment patterns

Repeated unwanted contact, obsessive follow-ups, and escalating pressure campaigns spread across multiple messages.

Factual content preserved

Times, dates, logistics, genuine questions, and practical requests pass through intact, so you still get the information you need.

Nothing is deleted

Even fully hostile messages that contain no usable content are archived in full. The record is always complete.

A filter so advanced you can just set and forget.

FenceChat’s filter is smart enough to know the difference between blunt and cruel. It doesn’t just catch swearing: it’s tuned to spot guilt-tripping, sarcasm, passive-aggressive language and more.

Trained on years of conflict research.

FenceChat is grounded in published research on how hostility actually shows up in language: the Four Horsemen of relationship breakdown, DARVO, coercive control, and workplace incivility. Not a keyword list.

Abuse detected and tagged.

The Legal plan doesn’t stop at filtering and archiving. It tags the pattern in each message, coercive or controlling language, financial abuse, and more, organised into a record for your solicitor.

We focussed on the filter.

Some co-parenting apps include message filtering as one feature alongside calendars, expense tracking and more. FenceChat does one job: filtering the messages you receive.

How the filter works, step by step

1

Someone messages your FenceChat address

They email or text as normal, with nothing to install. The first time, they get a one-time notice that messages are processed by AI and stored. Works for both email and SMS.

2

We receive it in full

The whole message reaches FenceChat first, and emails are saved, encrypted, the moment they land. Nothing is read until it has fully arrived.

3

Then it is processed by AI

It removes the hostility and keeps the useful information, reading for context, not just keywords. More on AI message filtering.

4

Both versions are kept

The untouched original and the filtered version sit side by side in your Message Archive. On the Legal plan, you can export it all as a PDF.

5

The clean version comes to you

It lands in your real inbox, forwarded from your FenceChat address. Your real email and number stay private. You can choose to have SMS messages forwarded to your phone too.

Frequently asked questions

Does the sender know their message was filtered?
The first time they message, they get a one-time notice that messages are processed by AI and stored. You can switch that notice off where telling them could put you at risk. Replies you send through FenceChat still end with a short “Sent via FenceChat using AI” line. After that, individual messages are not flagged to them, they do not see what was removed, and they never see your real email address or number.
What if a message is completely hostile with no useful content?
You will receive a short notification that a message arrived but contained only hostile content. The original is archived in full. You can view it at any time in your Dashboard without it landing directly in your inbox.
Will the filter catch subtle or indirect hostility?
The AI is designed to detect context and tone, not just explicit language. Passive-aggressive phrasing, loaded implications, and coded pressure are flagged where possible. No filter is perfect at edge cases, but the original is always preserved regardless.
Can I still access the unfiltered originals?
Yes. Every original is stored in your archive and accessible from your Dashboard at any time. On the Legal plan, you can export originals as a timestamped PDF suitable for legal proceedings.
Does filtering work for SMS as well as email?
Yes. Email filtering is included on all plans from A$14/month. SMS filtering is included on the Protected plan (A$29/month) and Legal plan (A$59/month), which include your FenceChat SMS number. Only contacts you add can text it, and filtered texts are delivered to your inbox and Dashboard.

Stop absorbing the abuse. Start using the filter.

Set up your FenceChat address in under two minutes. FREE 30-day trial.

FenceChat
See Plans and Pricing →

Essential plan from A$14/month. FREE 30-day trial.