With Latitude launching a new filter—and this update actually changing something unlike the last time—Filterspotter will no longer work usefully. If you still want to use it for some reason, you can find it here.

The new filter appears to be a Bayesian or AI classifier, which makes it much harder to characterize than the previous keyword filter. Latitude themselves likely don't know how it really works. The filter message has also been changed to "Uh oh, the AI doesn't know how to handle this situation," which is an outright lie.

I've recieved reports of new false positives, particularly around words suggesting incest, like "mom", and seen these in my own testing. However, I could not say whether this is intentional behavior or a side effect of their training data.

For reference, the rules of the last filter as of my latest knowledge are below:

Column A ("age-related") terms:

But not for some reason:

Column B ("sex-related") terms:

Tripped the filter alone: