Browser extension · Available for Chrome
The extension reads AI replies as you chat and frames possible patterns with a label. You decide what to do.
No account · Chat stays on your device · Arabic and English
HAI Lab
2 flags in 3 replies
01 · Pleasing, not honest
Read →
Changed its opinion after you pushed back, with no new reason.
02 · Keeps you hooked
Read →
Suggested a new question to keep you going.
Chat content stays on this device
Read
Reads the reply on your device and sends it nowhere.
Flag
Frames the questionable part and names the pattern.
Decide
Ask for a neutral answer, dismiss it, or send it as an example.
In plain words. Details are in the extension privacy policy.
On your device
Your chat text is checked on your device and is only sent if you choose to send it as an example.
No account
We do not need your name or email.
Your choice
Sending is optional, and personal data is removed first.
Few permissions
We ask only for what is needed and explain each one.
Six of the site’s patterns.
praise.unearned
hook.continue
request.altered
vendor.self
affect.simulated
risk.unreferred
Which sites does it work on?
It works on popular AI assistants in the browser, such as ChatGPT, Gemini and Copilot. Being on this list is about compatibility, not a judgement of any platform.
Is it always right?
No. It gives a flag with a confidence score, and you decide.
What it can and can't do
It runs in Chrome. Flags are possible patterns, not verdicts, and some will be wrong. We have not published accuracy figures yet. If a flag looks wrong, send it to us.
How do I turn it off?
From the extension button, or pause it for one site.