A message is blocked when its score reaches the threshold.
Square = harmful, circle = harmless, ring = a mistake, bold outline = the one you picked. Tap a mark to read it.
Read another message:
A toy: made-up messages that ask for or share contact details. Real filters score many kinds of harm, each against its own threshold.
The counts and the sentence then show 10,000 messages, 100 of them harmful, using the miss rate and false-alarm rate this filter has on the test messages. They are expected counts, not measured ones. The marks still show the test messages. 0 of 80 harmless messages blocked is not a rate of 0 in real traffic.