Skip to content
Log in Get our score
☢️ AI toxicity test for chats · doesn't take sides

Are we toxic? Upload your chat and get your relationship's toxicity score.

WhatsApp, Instagram or screenshots. Get a 0-100% toxicity score, the red flags with receipts, and each person's own score.

Get our toxicity score
Your score is freeChat deleted after scanThey never know
Chat ready? Skip the quiz and upload it now

Works with the chats you actually have

💬 WhatsApp export (.txt / .zip)📸 Instagram DMs (JSON / ZIP)💙 iMessage (screenshots, or a Mac/PC export)✈️ Telegram, Messenger, Snapchat, Android SMS🖼️ Screenshots: Hinge, Tinder, any app✍️ Paste text
Try it now

Score a sample chat

Pick one of our fictional chats and watch it get scored. Instant, no signup, nothing uploaded.

Fictional sample chats we wrote. The real check reads your whole chat with AI and counts the stats from every message.

L
Leocrush · sample chat
ready to scan
The toxicity score

What your score means

One score for the chat, plus one for each of you. It rates patterns, not people, and normal arguments don't count against you.

0-20%💚 Healthy

Disagreements happen and get repaired. Respect is the default.

21-40%🌶️ A little spicy

Some friction or bad habits, nothing that defines you. Easy wins.

41-60%😬 Kinda toxic

A loop is forming: guilt, jealousy or cold shoulders, more than once.

61-80%🚩 Toxic

Unhealthy patterns are the norm, often from both sides. Worth a real talk.

81-100%☢️ Very toxic

Contempt, control or cruelty run the chat. Put your wellbeing first.

🚩 Pushes the score up

  • Contempt, insults, eye-roll texts
  • Guilt trips and blame
  • Silent treatment, stonewalling
  • Jealousy or checking up on someone
  • Gaslighting, rewriting what happened
  • Hot and cold, ignoring boundaries

💚 Doesn't count against you

  • Slow replies or a busy week
  • Short texts from a short texter
  • Disagreeing, even loudly, then repairing
  • One bad night in a good chat
  • Saying no, setting a boundary

If a chat shows threats or abuse, there is no score. You get support resources first.

Sound familiar?

Built for the 2am "wait… are we toxic?"

"no reason. have fun" …is that a red flag or am I overthinking?
I always text first. Do they even like me?
They said "nothing serious" but get jealous. Which one is it?
Every fight feels like the same fight. Is that normal or toxic?
My mom's texts leave me feeling guilty and I can't say why.
Be honest… is it them, or is it me?

Your friends love you, which is exactly why they can't be objective. AreWeToxic reads the whole thing, scores it, shows both of your patterns, and tells you straight.

Sound like your chat?Stop guessing. Get your toxicity score, with receipts.
Get our score →
For an ex or a situationship

Will they come back?

Add a chat with an ex or a situationship and your report adds a playful read of the signals: who reaches out after the silence, the "random but I miss you" texts, the questions nobody answered, and anything that says it's really over.

  • A 0-100% "come back" read, with the reasoning
  • Signals for and against, quoted from your chat
  • Honest when the door is closed. Never "how to win them back"
Get our score →
How it works

From chat to toxicity score in about a minute

No bots join your chat. Nobody gets notified. Scroll through exactly what happens.

  1. 1

    Drop the chat

    Export from WhatsApp or Instagram, add 1-10 screenshots from any app, or paste the text. Exporting never notifies the other person.

    • WhatsApp .txt or .zip
    • Instagram JSON or the full ZIP
    • Screenshots: iMessage, Snap, dating apps
  2. 2

    Tell us which one is you

    We show who's in the chat and the date range, and you tap your name. That way each person gets their own score, and your suggested replies are written for you.

    • Big exports with many chats: pick the one you mean
    • Your chat is held max 20 minutes while you confirm
  3. 3

    The toxicity scan

    Code counts every message: who starts conversations, reply times, follow-ups. Then AI reads for guilt trips, jealousy, silent treatments, contempt and the good stuff too.

    • Stats from all messages, never guessed
    • Long chats: opening, latest stretch and key moments
  4. 4

    Your toxicity score, with receipts

    A 0-100% score for the chat, one for each of you, and the red and green flags behind it. Every flag quotes the exact message, and we check each quote word-for-word against your chat.

    • Mixed signals and contradictions
    • What each of you could own
    • Attachment hints, never diagnoses
  5. 5

    Know what to say next

    Two or three replies tuned to your goal (clarity, closure or confidence), ready to copy. Not sure about your own draft? Run it through the Reply Coach.

    • The raw chat is deleted after the scan
    • Delete the report anytime, in one tap
    Get our score →
What we look for

What goes into the score

Receipts, not vibes. Each signal is shown with the numbers or the exact messages behind it.

Effort split

You 64Them 36

One score from who starts, words, questions and reply speed. 50/50 means matched.

Reply time

You: median 4 minThem: 1.3 h

Median, not average, so one slow day doesn't skew it.

Who starts

You started 9Them 3

A new conversation = first message after 4+ hours of quiet.

Double texts

hey stranger 👀saw your story!2nd text, no reply in 1h+

Follow-ups sent after an hour with no answer.

Message length

so the meeting went really well and then my boss said…cool

Average words per message for each person. Dry texter, or just brief?

Questions asked

You asked 11They asked 5

Curiosity is one of the clearest interest signals in text.

Late-night texts

Messages between midnight and 5am. Context matters, so we flag patterns, not single nights.

Red & green flags

Them ✓ verifiedyou could have at least said you got home

Each flag cites the exact message. ✓ means we found it word-for-word.

Mixed signals & contradictions

SAIDnot looking for anything seriousBUTwhy didn't you text me back tonight

What was said vs. what was done, side by side.

Play with the math

How the effort score works

Effort sits next to your toxicity score in the report. Move the sliders: this is the same formula, computed from every message.

Weights: conversation starts 30%, words 25%, questions 20%, reply speed 25%. Both scores always add up to 100.

You0
Them0
Balanced

You're both showing up.

Get our real scores →
Reply speed, explained

What a slow reply does (and doesn't) mean

Reply time is the most over-read signal in texting. Here's how to read it like a pro.

Why we use the median, not the average

median: 15 min average: 1 h 13 min one 9 h reply

One busy afternoon can drag an average way up. The median is the middle reply, so a single slow day can't distort it.

Compare to their own baseline

Someone who always replies in hours isn't pulling away when they reply in hours. A change is the signal, not the number.

Context explains a lot

Work shifts, school, kids, time zones and "phone on silent" people all look slow on paper.

The trend beats the moment

One slow week is noise. Replies getting slower month after month, while effort elsewhere drops too, is a pattern.

Speed isn't interest

Fast, short replies can be low effort; slow, thoughtful ones can be high effort. That's why speed is only 25% of the score.

How AreWeToxic measures it

  • Reply time: the gap between the other person's last message and your first reply. Gaps over 3 days are left out.
  • Conversation start: the first message after 4+ hours of quiet.
  • Unanswered follow-up: a second message sent an hour or more after your last one, with no reply in between.
  • Late-night message: sent between midnight and 5am.
  • Screenshots: no timestamps, so reply-time stats are skipped instead of guessed.
Impartial by design

We don't take your side. We take the truth's.

A score that only blames them feels great for five minutes and changes nothing. AreWeToxic scores each person separately and gives everyone a mirror, you included.

  • ✓A score for each of youNot a villain and a victim. Two people with patterns, each with their own number.
  • ✓"Could own" and "does well" for each personWhat each of you could change, and what each of you does well.
  • ✓You choose the deliveryBrutal, honest-but-kind or gentle. The facts don't change.
  • ✓No blame when it isn't fairIf a chat shows threats or control, we never ask the person being hurt to "own their part".
SAMPLE · fictional chat

⚖️ Both sides

Maya you Could own
  • Said "me neither" to casual when she wanted more
  • "it's fine whatever" instead of saying what hurt
Does well
  • Asks directly what this is
Jordan Could own
  • Wants check-ins without commitment
  • Goes quiet, then expects instant replies
Does well
  • Remembers what matters to her
  • Honest about being scared
Example reports

Sample toxicity reports

Five sample reports built from fictional chats we wrote. Yours will be specific to your messages.

Get our toxicity score →
More sample snippets

What AreWeToxic notices, chat by chat

Real report formats, built from fictional chats we wrote. Filter by relationship.

Partner

Apologies with a "but"

Alex ✓ verifiedsorry you felt that way but you know how busy I am

The apology shifts the problem onto how you felt. A cleaner repair names what they did.

Partner

Checks in after conflict

Sam ✓ verifiedhey, still thinking about earlier. can we talk tonight?

Coming back to a fight on purpose is one of the strongest green flags in long relationships.

Partner

Scorekeeping

You ✓ verifiedthat's the 4th time this month btw

Fair frustration, but counting turns the next talk into a trial. Both of you could own this one.

Crush

Remembers small details

Leo ✓ verifiedwait how did the audition go??

Following up on something you mentioned days ago is a real interest signal, even if replies are slow.

Crush

Future talk, no plan

Leo ✓ verifiedwe should def hang sometime

"Sometime" isn't a no. Offer two concrete days and see what happens.

Crush

Only texts after 11pm

Kai ✓ verifiedu up? 👀

When most contact is late at night and never turns into daytime plans, notice the pattern.

Situationship

Jealous but "nothing serious"

Jordan ✓ verifiedguys from class?

Wanting exclusivity without offering it is a double standard worth naming.

Situationship

You agreed to casual too

You ✓ verifiedoh totally, me neither, no pressure 🙂

Impartial means this counts too: hoping for more while saying "me neither" sends a mixed signal of its own.

Situationship

Honest about fear

Jordan ✓ verifiedi'm just scared of messing it up

Naming a feeling is a big step for avoidant texters. A good moment to ask what this is.

Ex

"Miss you" with no follow-through

Chris ✓ verifiedrandom but i miss talking to you

Nostalgia texts reopen the wound without offering anything new. Decide what you want before you reply.

Ex

Takes responsibility

Chris ✓ verifiedi know i made it hard at the end

Accountability makes closure possible, whether or not you reconnect.

Ex

You reply within a minute

StatsYour median reply: 1 min · Theirs: 9 h

Your speed says you're still invested. Useful to know before deciding on contact.

Friend

Last-minute cancels

Nora ✓ verifiedomg so sorry can we raincheck

Three in a month is a pattern. Talk about it before resentment does the talking.

Friend

Shows up when it counts

Nora ✓ verifiedon my way with ice cream, don't argue

Low effort day to day, high effort in a crisis. Both are true at once.

Friend

You always plan

StatsYou started 74% of conversations

Not a verdict, a conversation starter: ask them to pick the next plan.

Family

Guilt as a request

Mom ✓ verifiedI guess you're too busy for your mother

The need underneath is "I miss you". Answering the need, not the guilt, usually works better.

Family

Practical love

Mom ✓ verifieddid you eat? I left soup in your freezer

Care shown through doing. Worth naming back: "the soup was perfect, thank you".

Family

Short replies when hurt

You ✓ verifiedmom.

Shutting down is understandable, but it reads as cold. One honest sentence lands better.

Work

Urgency after hours

Dana ✓ verifiedneed this by 8am, thx

Sent at 11:40pm. A recurring pattern worth raising calmly, with examples.

Work

Credit given in public

Dana ✓ verifiedshoutout to Ravi for saving this launch

Recognition is a real green flag in work chats. Keep it in mind next time you push back.

Work

Passive "per my last message"

You ✓ verifiedas I mentioned twice already...

Understandable, but it escalates. The Reply Coach can make it firm without the edge.

Every snippet above is fictional. Your report quotes your own messages and marks each quote ✓ verified or paraphrased.

Inside the report

Anatomy of a toxicity report

A real report screen from AreWeToxic, generated from a fictional chat we wrote. Tap a number to see what it is.

AreWeToxic report screen: a 28% toxicity score labelled A little spicy, each person's own score (Maya 12%, Jordan 34%), the verdict, a share button, and a red flag with two verified quotes
  1. Your toxicity score0-100% for the whole chat, judged on how often and how badly unhealthy patterns show up. Normal arguments don't count against you.
  2. The band and the flag countHealthy, a little spicy, kinda toxic, toxic or very toxic, plus how many red and green flags we found.
  3. Each person's own scoreBoth of you get a number and your main pattern, so it never turns into a villain and a victim.
  4. The verdictOne fair sentence, in brutal, honest or gentle delivery. Then share your score card (arewetoxic.com) if you want.
  5. Red flags with severityWho they're about and why they matter. Below: green flags, effort stats, both sides and what to say next.
  6. Verified quotesEach quote is searched in your chat. Found word-for-word gets a ✓; otherwise it's labelled as paraphrased.
That was a sample. Yours will quote your messages.1-minute quiz, then upload. The chat is deleted after the scan.
Get our score →
Texting signs library

16 texting signs, decoded honestly

What each sign often means, and what it doesn't prove. Tap a card to open it.

↗They ask follow-up questions

Often means: genuine curiosity about your life.

Doesn't prove: romantic interest. Friendly people ask too. Look for questions plus plans.

↗They remember small details

Often means: they listen and you're on their mind between chats.

Doesn't prove: commitment. It's a good sign, not a promise.

↗They start conversations too

Often means: they think of you unprompted. A start ratio near 50/50 is healthy.

Doesn't prove: anything on its own if the conversations stay shallow.

↗They suggest specific plans

Often means: real intent. "Thursday at 7?" beats "we should hang sometime".

Doesn't prove: exclusivity. Ask, don't assume.

↘Replies are getting slower

Often means: a busier season, or lower priority. The trend matters more than one day.

Doesn't prove: they've lost interest. Compare to their own baseline.

↘Messages got shorter

Often means: less energy for the chat right now.

Doesn't prove: anything if they've always been a brief texter.

↘You start almost every chat

Often means: the effort is lopsided, at least over text.

Doesn't prove: they don't care. Some people never text first. Ask them to plan the next one.

↘Plans keep "falling through"

Often means: low priority, or something going on they haven't shared.

Doesn't prove: bad intent. Ask directly once, then watch the next two weeks.

!"It's fine" when it isn't

Often means: hurt that feels risky to say out loud.

Doesn't prove: they're done. It's an invitation to ask gently.

!Silence after a disagreement

Often means: cooling off, or avoiding the conflict.

Doesn't prove: punishment, unless it repeats every time you set a boundary.

!"Always" and "never"

Often means: built-up frustration about a real pattern.

Doesn't prove: the pattern is that absolute. Swap in a specific example.

!Sarcasm in serious moments

Often means: discomfort with vulnerability.

Doesn't prove: contempt, unless it's a habit aimed at you.

♥"Did you get home ok?"

Often means: care, as a one-off.

Doesn't prove: care if it becomes a required check-in. Then it's control.

♥They apologize specifically

Often means: real accountability: "I was short with you, that wasn't fair".

Doesn't prove: change. Watch what happens next time.

♥They come back to a fight

Often means: the relationship matters more than being right.

Doesn't prove: agreement, and that's fine. Repair beats agreement.

♥They celebrate your wins

Often means: they're on your side.

Doesn't prove: anything if it never shows up in hard moments. Look for both.

Read the full guide: Is he losing interest? 12 texting signs →

Red flag vs green flag

The flag encyclopedia

Most red flags have a healthy twin. Same topic, very different energy.

🚩 Red"Who are you with?" as a demand

Required check-ins and proof.

💚 Green"Have fun! Text me if you need a ride"

Interest without control.

🚩 Red"Sorry you feel that way"

An apology for your feelings, not their actions.

💚 Green"I was wrong to say that"

Names the action, owns it.

🚩 RedDays of silence after a no

Saying no gets punished.

💚 Green"I need tonight, let's talk tomorrow"

Space with a return date.

🚩 RedJokes that are digs

Teasing only one person laughs at.

💚 GreenInside jokes you both start

Playfulness that goes both ways.

🚩 RedGuilt instead of asking

"I guess I'm not a priority."

💚 GreenA direct ask

"I miss you. Sunday?"

🚩 Red"I never said that"

When it's right there in the chat.

💚 Green"You're right, I did say that"

Memory slips get owned.

🚩 RedLove-bombing in week one

"You're my person" before they know you.

💚 GreenSteady warmth over time

Consistent, not intense.

🚩 Red"Your friends don't get us"

Nudges that shrink your world.

💚 Green"Bring your friends!"

Wants to know your people.

🚩 RedAsking again after a no

Photos, plans or favors, five times over.

💚 Green"No worries!"

A no is accepted the first time.

🚩 RedHot, then ice cold

Intense days, then nothing, on repeat.

💚 Green"Busy week, thinking of you"

Busy, but still in touch.

🚩 RedWants passwords "for trust"

Surveillance dressed as closeness.

💚 GreenTrust by default

Asks questions, accepts answers.

🚩 RedThreats, even "dramatic" ones

"You'll regret this."

💚 GreenDisagreeing safely

Angry is allowed; scary is not.

Read: 12 red flags worth taking seriously (and 5 that usually aren't) →

Mixed signals, summarized

When words and actions don't match

Mixed signals are confusing because both signals are usually real. Five patterns, and what to do.

1

Casual label, committed behavior

"Nothing serious", then jealousy. Often: they like you more than they admit, or want comfort without risk.

2

Hot and cold

Warm for days, then silence. Look at what happened right before each cold spell.

3

Great in person, dry in text

Some people hate texting. Judge by plans kept, not message length.

4

Big words, small follow-through

"Let's travel together!" and no plan, ever. Ask for one small, concrete thing.

5

1am intimacy, noon distance

Deep at night, polite by day. Sweet or convenient: watch whether it becomes daytime plans.

+

And your own signals

"Me neither" to casual while hoping for more is a mixed signal too. Fixing yours is the fastest win.

What to do: name the pattern, not the person. Ask one clear question. Say what you want. Then watch the next two weeks.
"I like you, and I'm confused about what this is. Can we talk about what we both want?"

Read the full guide on mixed signals →

Texting glossary

The words for what's happening

Naming a pattern makes it easier to talk about. Tap any term.

Breadcrumbing

Small bits of attention (a like, a "haha") that keep you hoping, with no real investment behind them.

Ghosting

Ending contact with no explanation. Different from a slow reply: ghosting doesn't come back.

Orbiting

They stopped talking to you but still watch your stories and like your posts.

Love-bombing

Overwhelming affection very early, often followed by withdrawal or control once you're attached.

Situationship

More than friends, less than defined. Fine if both want it; painful if one wants more.

Double text

A second message before they replied. Normal now and then; a pattern if you're always the one doing it.

Dry texting

Short, low-energy replies ("ok", "lol"). Can be a style, a mood or disinterest. Compare to their baseline.

Left on read

Seen, not answered. Often means "later", sometimes means "no". Context and patterns decide.

Gaslighting

Repeatedly denying things that happened so you doubt your memory. The chat is a record: scroll back.

Stonewalling

Shutting down completely during conflict. A cool-off with a return date is healthy; a wall isn't.

Zombieing

A ghost comes back as if nothing happened. Decide what you want before you reply.

Benching

Keeping you "on the bench" with occasional contact while they keep options open.

Soft launch

Hinting at a relationship (a hand in a photo) before making it official.

Repair attempt

Any move to de-escalate a fight: a joke, an apology, "can we start over?". A strong green flag.

Attachment styles, gently

How attachment can show up in texts

Hints, never diagnoses. Texts show only part of a person, and most people mix styles depending on the relationship.

Secure-leaning

  • Says what they want, plainly
  • Slow replies don't trigger panic
  • Comes back to fights to repair

Anxious-leaning

  • Fast replies, frequent follow-ups
  • Reads silence as rejection
  • Seeks reassurance after small changes

Avoidant-leaning

  • Pulls back when things get close
  • Keeps labels vague
  • Prefers logistics over feelings

Mixed / fearful

  • Wants closeness, then flees it
  • Hot and cold cycles
  • Big feelings, hard to say plainly

In a report, attachment notes look like "may lean anxious, because..." with a caveat. They're a lens for curiosity, not a label for anyone.

Recognized a sign? See if it's a pattern.Flags with the exact messages, for both of you.
Check our chat →
Included in every plan

Three more tools for the in-between moments

Not every question needs a full toxicity check.

Text Decoder

Paste one confusing message. Get the likely meanings (not just the scariest one) and replies.

"k. have fun i guess"
likelyA little disappointed, covering it with a shrug
possibleJust low energy tonight
less likelyPulling back for good

Reply Coach

Draft your reply, see how it may land, and get clearer, warmer or firmer versions before you hit send.

"whatever, do what you want"
high riskReads as a door closing
I get you were worried. Can we agree on what this is before check-ins become a thing?

Portrait

Combine 2-5 checked chats to see how you show up across relationships: what you repeat, where you shine.

Your facesPatient initiator with crushes, peacemaker with family
You repeat"It's fine" before saying what's wrong
Try thisLet them start the next two conversations
What to say

Scripts for the hardest texts

Starting points for the messages people draft and delete ten times. Your report writes ones specific to your chat.

Ask for clarity
I like where this is going and I'd rather not guess. What are you looking for with us?
Set a boundary
I'm happy to tell you about my night, but I'm not okay with being asked to check in. Can we talk about trust?
Repair after a fight
I've been thinking about earlier. I was short with you and that wasn't fair. Can we try again tonight?
Turn vague into a plan
I'd love to. Thursday or Saturday?
End it kindly
I've really valued our time, and I've realized we want different things. I'm going to step back. I wish you well.
Answer guilt with love
I love you and I want to visit. Sunday works. When it's framed as me being too busy, it makes it harder, not easier.
Ask a friend for more
I love our hangouts. Could you pick the next plan? I'd love to be invited for once.
Reply to an ex
I'm glad we can be kind to each other. I need some space to move on, so I'm going to stop texting for a while.

Drafting your own? The Reply Coach shows how it may land and offers clearer, warmer or firmer versions.

Portrait, in depth

Your patterns across 2-5 chats

Included in every plan. Combine your checked chats to see how you show up with different people.

01

The headline

One line about how you relate, built from all your chats.

02

Your faces

How you show up with a partner vs. a parent vs. a friend.

03

What you repeat

The move you make with everyone, like "it's fine" before saying what's wrong.

04

Your contradictions

Direct with friends, indirect with crushes? That's useful to know.

05

Strengths and growth edges

What to keep doing, and the one thing worth practicing.

06

One thing to try

A concrete experiment for next week, like letting them start the next two chats.

Chats in safety mode are never mixed into a Portrait.

How people use it

Four illustrative stories

Composite, fictional stories that show typical ways people use AreWeToxic. Not real users or reviews.

Illustrative story · fictional

Maya and the "nothing serious" guy

  1. BeforeRe-read the chat every night, unsure if she was overreacting.
  2. The score28% "a little spicy": Jordan 34%, Maya 12%. "Territorial double standards" with verified quotes. Also: her own "me neither".
  3. AfterAsked one clear question over coffee instead of another 1am spiral.
Illustrative story · fictional

Dev and the friend who always cancels

  1. BeforeQuietly resentful, about to drift away.
  2. The score27% "a little spicy". Dev started 74% of chats, but the friend showed up in every crisis.
  3. AfterSaid "could you plan the next one?" The friend did.
Illustrative story · fictional

Ana and the guilt-trip texts

  1. BeforeDreaded every message from her mom.
  2. The score46% "kinda toxic": guilt as a request, plus constant practical care. And Ana's own one-word replies.
  3. AfterAnswered the need instead of the guilt, with a Sunday plan.
Illustrative story · fictional

Kai and the ex who texts at 1am

  1. BeforeEvery "miss you" restarted the heartbreak.
  2. The score39% "a little spicy": late-night contact, no daytime follow-through, and Kai replying within a minute.
  3. AfterSent a kind, clear goodbye and muted the chat.
Who it's for

Any chat that makes you ask "are we toxic?"

🦋The talking stageAre they into you, or just bored?
🌀SituationshipsMake sense of hot-and-cold.
💞CouplesSee the argument loop from outside.
💔ExesClosure, with receipts.
🫶FriendshipsIs it one-sided, or just a busy season?
🏡FamilyName the guilt-trip, keep the love.
💼WorkRead a tricky coworker or client thread calmly.
👥Group chatsEffort stats for up to 8 people.
🪞YourselfYour patterns across chats, with Portrait.
After your score

What changes when you can see the pattern

Not a magic answer. A clearer head, and a better next message.

Stop re-reading at 2am

The questions that loop in your head get one calm, outside answer you can come back to.

Know what to send

Replies matched to your goal, so you stop drafting and deleting the same message ten times.

Talk with receipts, not vibes

Point to the exact messages, calmly, instead of "you always" and "you never".

See your own part

Impartial means you get a mirror too. That's usually where the easiest fix is.

Decide calmly: fix, wait or go

A straight verdict and the pattern behind it, so the decision is yours and not your anxiety's.

Get support if it isn't safe

If a chat shows threats or control, you get resources and a calm read, not a score.

Myth vs fact

Texting myths that cause a lot of 2am spirals

Tap a card to flip it.

Facts beat spirals.Get your chat's toxicity score in about a minute.
Get our score →
Export guide

Get your chat out in 20 seconds

Pick where you talk. Exporting never notifies the other person.

Full walkthroughs: WhatsApp guide · Instagram guide

Why AreWeToxic

How it compares

Against another chat-reading app, a general chatbot, and your group chat. Fifteen things that matter.

Swipe to compare →
FeatureAreWeToxicRayan KnowsChatGPT copy-pasteAsking friends
Toxicity score (0-100%), one per person✓Not listedIf you askVibes only
WhatsApp + Instagram exports✓✓Paste only✕
Screenshots (iMessage, Snap, apps)✓✕Manual✓
Effort stats computed from timestamps✓Not listed✕✕
Quotes checked against your chat✓Not listed✕✕
Both sides, by design✓✓If you askRarely
Text Decoder + Reply Coach✓✕ManualChaotic
Multi-chat PortraitIncluded$18.99 each✕✕
Price per chat reportScore free, full report in the free week$7.49SubscriptionFree
"Will they come back?" read for exes✓Not listedIf you askEndless debate
Safety mode with resources✓Not listedVariesMaybe
Free instant demo on the homepage✓Free preview✓✓
Group chatsUp to 8 people✓Paste onlyMessy
Choose the tone (brutal, honest, gentle)✓Not listedIf you askDepends on the friend
Report deliveryIn-app, instantlyPDF by emailIn the chatWhenever they reply
Saved history of reports✓Not listedChat history✕
Attachment hints with caveats✓Not listedIf you ask✕
Relationship typesPartner to coworkerLove, crew, friends, family, workAnyAny
Raw chat deleted by default✓Not listedPer settingsScreenshots live forever

Rayan Knows details as listed on rayanknows.com, Oct 2026. "Not listed" means we couldn't find it on their site. ChatGPT is a trademark of OpenAI; AreWeToxic is not affiliated.

Honest comparison

Which help fits which moment?

AreWeToxic isn't the answer to everything. Here's when each option shines.

AreWeToxic

Best for: "Are we toxic, and who's doing what?"

  • Reads the whole chat, scores it and each of you
  • Quotes checked against your messages
  • Both sides, plus what to say next
  • Not therapy, and can't see what happens offline

A therapist

Best for: patterns that go deeper than one chat

  • Knows your history and context
  • Can help with trauma, anxiety, safety planning
  • Weekly sessions, higher cost
  • Won't read 3,000 messages for you

Your best friend

Best for: comfort and loyalty

  • Knows you, takes your side
  • Free, and they care
  • Sees only the screenshots you chose
  • Loyalty isn't objectivity

A general chatbot

Best for: quick questions about one message

  • Fast and flexible
  • Paste-only, stats guessed not counted
  • No quote checking, no saved reports
  • Tends to agree with how you frame it

Journaling

Best for: knowing what you feel

  • Private, calming, builds self-awareness
  • Only your side of the story
  • No outside read of the other person

Support lines

Best for: when you don't feel safe

  • Trained people, often 24/7
  • Safety planning and real resources
  • Use them first if there are threats or control
Privacy deep-dive

Your chat is not our business model

Here's exactly where your messages go, and where they don't.

1. You uploadOver an encrypted connection
2. We parse & countHeld temporarily (max 20 min) while you confirm
3. AI writes the reportSent to our AI provider only for this
4. Raw chat deletedRight after your report is built. We never store it.
5. Report saved to youDelete it anytime, in one tap

✓ What we keep

  • •Your report: stats, flags, short evidence quotes, suggestions
  • •Your email and a hashed password
  • •Your quiz answers, to personalize the app
  • •A free score you don't unlock: 7 days, then it's deleted

✕ What we don't

  • •We don't keep raw chats: they're deleted right after your report is built (max 20 min)
  • •We don't sell data or show ads
  • •We don't train models on your chats
  • •We never contact the people in your chat

🫂 Safety mode

If a chat shows threats, coercive control or self-harm, AreWeToxic switches tone: support resources first, no scores, no "flag counting", and no blame on the person being hurt. If you're in danger right now, call your local emergency number.

Privacy, in detail

Every piece of data, where it goes, and for how long

No fine print. If it isn't in this table, we don't collect it.

DataWhyKept for
Raw chat (file, screenshots, paste)To count stats and write your reportMax 20 minutes, deleted right after the scan. Never stored.
Your reportSo you can reread it and build PortraitsUntil you delete it (one tap, or "delete all"). A free score you never unlock: 7 days.
Short evidence quotesTo show the receipts behind each flagPart of the report, deleted with it
Email + hashed passwordYour accountUntil you close your account
Quiz answersTo personalize tone and focusWith your account
Text sent to our AI providerOnly to generate your reportNot used to train our models; never sold
Encrypted in transitUploads travel over HTTPS.
Auto-deleteAbandoned uploads are swept after 20 minutes.
They never knowNo notifications, no bots, no contact.
No ads, no data salesWe're paid by subscriptions, not by your data.

Read the full privacy policy →

Safety first

When it isn't a red flag, it's a danger

Threats, monitoring, pressure for photos, or threats of self-harm used to keep you from leaving are not "texting styles". If AreWeToxic sees them, it switches to safety mode: support resources first, no toxicity score, and never blame on the person being hurt.

Tip: if someone checks your phone, delete your report after reading it. One tap.

Before you start

A 1-minute quiz tunes your toxicity check

A few quick taps. Your answers set the focus and the tone of your report.

Quick questionWho is this chat with?
💞 Partner🌀 Situationship🫶 Friend💔 Ex
Quick questionWhat feels off?
JealousySilent treatmentHonestly, might be me
Quick questionHow do you want the truth delivered?
🌶️ Brutal🤝 Honest but kind🧸 Gentle
Start the quiz →
FAQ

Your questions, answered

Grouped by topic. Still unsure? Email us.

The toxicity score

How is the toxicity score calculated?
The AI reads your chat against a fixed list of unhealthy patterns: contempt or insults, guilt-tripping, stonewalling or the silent treatment, jealousy or control, gaslighting, hot-and-cold behavior, defensiveness and blame, passive aggression and ignoring boundaries. It weighs how often and how badly they show up, then places the dynamic on a 0-100 scale with fixed bands. Each person also gets their own score. The effort split (who starts, words, questions, reply speed) is counted in code from every message.
What does 73% toxic mean?
It falls in the Toxic band. The bands are: 0-20 Healthy 💚, 21-40 A little spicy 🌶️, 41-60 Kinda toxic 😬, 61-80 Toxic 🚩, 81-100 Very toxic ☢️. A score in the 60s or 70s means unhealthy patterns show up often enough to shape the chat, not just on one bad night. Your report lists the red flags behind it, with the exact messages.
Will it take my side?
No. AreWeToxic scores the dynamic, then gives each person their own score and pattern, you included. Every report shows what each of you could own and what each of you does well. You choose brutal, honest or gentle delivery, but the facts stay the same.
Can I check a chat with a friend, my mom or my ex?
Yes. It works for partners, situationships, crushes, friends, exes, family and group chats. You pick the relationship before the check so the report reads it in the right context.
Is a high score a reason to break up?
No. The score is a conversation starter, not a verdict on a person. Texts show only part of a relationship, and people can change patterns once they see them. Use the report to talk, set a boundary or get support, and trust what you know beyond the chat.
Do arguments or slow replies make a chat toxic?
Not on their own. Normal disagreements, busy days and slow replies are part of most chats. The score rises when unhealthy patterns repeat or hit hard, like insults, guilt trips or the silent treatment.
What does "✓ verified quote" mean?
We search your chat for each quote the AI cites as evidence for a red or green flag. If it appears word-for-word, it gets a ✓. If not, it's marked as paraphrased so you know.

Getting started

How do I export a WhatsApp chat?
Open the chat, tap the name (iPhone) or ⋮ then More (Android), then Export chat and choose Without media. Upload the .txt or .zip. It takes about 20 seconds.
How do I get my Instagram DMs?
Settings, Accounts Center, Your information and permissions, Download your information. Pick Messages and format JSON. Instagram emails you a ZIP, often within an hour. Upload the ZIP and pick the conversation.
What about iMessage, Snapchat, Telegram or dating apps?
iPhone has no export for iMessage, so the fastest way is 1-10 screenshots, oldest first: we read them and you confirm which side is you. With a Mac you can export the chat for free with imessage-exporter (TXT), or on a PC or Mac with iMazing (CSV). We also read Telegram Desktop (JSON or HTML), Snapchat My Data, Facebook Messenger, Android SMS backups (SMS Backup & Restore XML) and Discord copy-paste. The upload screen shows tap-by-tap steps for each app. Screenshots have no timestamps, so reply-time stats are skipped instead of guessed.
Can I just paste the text?
Yes. Paste the conversation, ideally with names like "Alex: hey", so we can tell who said what and give each person their own score.
How long does it take?
About 2 minutes in total: a 1-minute quiz, around 20 seconds to export, and 15-40 seconds for the toxicity check. Your score shows up right away. Instagram needs time to prepare its export first.
How long can a chat be?
Exports up to 60 MB. For very long chats, the stats cover every message, and the AI reads the opening, the most recent stretch and the key moments.
What languages does it support?
The report is written in the main language of your chat. The counted numbers work in any language; the four horsemen signals (criticism, contempt, apologies and so on) are counted in English and Spanish and left out for other languages rather than guessed.
Does it work for group chats?
Yes. Effort stats work for up to 8 people, each person gets their own score, and the report looks at the group dynamic.

Privacy and safety

Does the other person get notified?
No. Exporting a chat never notifies anyone, and AreWeToxic never contacts anyone. Screenshots are silent in WhatsApp, iMessage and normal Instagram DMs. The exception is Snapchat (and Instagram vanish mode): they tell the other person about screenshots, so for Snapchat use its My Data export.
What do you keep, and for how long?
Only the report (scores, the numbers we count, flags and a few short example quotes, suggestions) until you delete it. We never store the chat text: after upload, the parsed chat waits at most 20 minutes while you confirm which person is you, then it is deleted right after your report is built. A free score you never unlock is deleted automatically after 7 days.
Do you train AI on my chats?
No. Your chat is sent to our AI provider only to write your report. We don't sell data or run ads.
Can I delete my reports?
Yes. Delete any report in one tap, or everything at once from the account menu.
Is it okay to upload a chat with someone else in it?
Only upload conversations you're part of, and use the report for your own understanding. Never use it to harass, monitor or control anyone.
What if the chat shows abuse or threats?
AreWeToxic switches to a safety mode: support resources first, no toxicity score and no blame on the person being hurt. If you are in danger, call your local emergency number.

The report

What else is in the report besides the score?
Red flags and green flags with the exact messages, each person's score and main pattern, the numbers counted from every message (who texts first, reply times by time of day, balance, momentum, the four horsemen and their antidotes, unanswered questions), a short read of those numbers with things to try, what changed since your last check, mixed signals, contradictions, what each of you could own, and replies you could actually send next.
How is the effort split calculated?
In code, from all messages: who starts conversations (30%), words (25%), questions (20%) and reply speed (25%). Everyone's shares add up to 100, so 50/50 means perfectly matched.
What is "Will they come back?"
For a chat with an ex or a situationship, your report adds a playful 0-100% read of whether they will reach out again (or stick around), with the signals for and against quoted from your chat. It is a read of signals, not a prediction, and it never coaches you to chase anyone.
What are attachment hints?
Possibilities like "may lean anxious", always with a reason and a caveat. They're a lens for curiosity, never a diagnosis.
Is this therapy or a diagnosis?
No. It's an outside read of text messages, which only show part of a relationship. For deeper patterns, a therapist is the right help.
Something in my report looks wrong. What can I do?
Check whether the quote is marked verified. Texts miss context like calls and time together, so trust what you know too. You can also delete the report and run a new check.

Tools and plans

What are the Decoder and Reply Coach?
The Decoder reads one confusing message and gives likely meanings plus replies. The Reply Coach shows how your draft may land and offers clearer, warmer or firmer versions.
What is a Portrait?
Combine 2-5 of your toxicity reports to see how you show up across relationships: what you repeat, your contradictions, strengths and one thing to try. Included, no extra fee.
How is this different from pasting into a chatbot?
AreWeToxic reads full exports and screenshots, scores the dynamic on fixed bands, gives each person their own score, computes the effort stats in code, checks every quote against your chat, and saves your reports for Portraits.
Is the score really free?
Yes. Add your chat and you get your toxicity score and your biggest flag without signing up. Both of your scores, every flag, the effort split, "Will they come back?" and what to say next unlock with the free week.
How much does it cost? Can I cancel?
Your toxicity score and your biggest flag are free, no signup. The full report and every tool come with a 7-day free trial, then $9/week. Cancel anytime: tap the chat button in the app to message us, or use Manage subscription on your last free day. Cancel before day 7 and you pay nothing.
Is there a limit?
One free score a day without an account. During the free week you can check up to 5 chats a day, and up to 10 a day after that, which covers far more chats than most people ever need.
Do I need to install an app?
No. AreWeToxic works in your phone or computer browser.
Guides

Read the signs yourself

Free, honest guides on toxic texting patterns, mixed signals and getting your chats out.

Toxic or not?

Is my relationship toxic? 12 texting signs

What pushes a chat's score up, and what doesn't.

Read guide → Red flags

Red flags in texts: 12 worth taking seriously

And 5 that usually aren't.

Read guide → Texting signs

Is he losing interest? 12 texting signs

And what each one doesn't mean.

Read guide → Mixed signals

Mixed signals over text

What they usually mean and what to do.

Read guide → Exes

Will my ex come back?

9 signs in their texts, and what they don't prove.

Read guide → How-to

Export a WhatsApp chat

iPhone and Android, step by step.

Read guide → How-to

Download your Instagram DMs

The JSON export, explained.

Read guide →

So… are you toxic?

Take the 1-minute quiz, add your chat and get your toxicity score free. Both of you get a mirror.

Get our toxicity score →
Get our toxicity score →