How the Stricter AI Filter Works

A score from 0 to 100

When you send a chat message, TalkSafe uses AI to assess how hostile or hurtful the wording is towards the recipient. It assigns a toxicity score from 0 to 100: a lower score means less toxicity, and a higher score means more. It looks at the meaning and tone, including insults, belittling, sarcasm and threats, rather than just checking for swear words.

The score is an assessment of the message, not a judgement of you as a parent. You can disagree, express frustration and set boundaries. The filter allows some sharp or mildly toxic wording; it does not require every message to sound perfectly friendly.

Send, rewrite or stop

• Lower toxicity: the message can be sent without a warning.
• A warning and suggested rewrite: when the wording is more likely to fuel conflict, the sender gets a warning and a suggestion for a calmer version. You can use the suggestion, edit the message yourself, or still choose to send the original if it has only received a warning.
• Very toxic wording: the message is blocked completely. You cannot send that original version by dismissing the warning. You need to change the wording before sending.

The other parent does not receive a blocked message or see the rewrite suggestion shown to you. They see the version that is actually sent.

What changes with the stricter filter?

Turning on the stricter AI filter in Settings lowers both the warning and blocking thresholds. Wording that would pass at the standard level may trigger a warning, and wording that would only get a warning may be blocked. This gives the chat less tolerance for toxic language.

The stricter level applies in both directions: to messages you send and messages you receive. It is enough for one participant in the chat to enable it. Turning your own setting off does not disable TalkSafe, and the stricter level still applies if another participant has it enabled.

Your setting is private

The other parent cannot see whether you have turned the stricter filter on or off. Your chosen setting is not displayed to them, and the app does not identify who enabled the stricter level.

Does the Co-parenting team read my messages?

The assessment is automated. The AI processes the text to assess its tone and suggest a rewrite. If you contact us about a message you believe was wrongly blocked, we can look into it. You can read more in our privacy policy.

AI can misunderstand wording or context. If a message is flagged, try making the practical request clearer and removing personal remarks. You can also read our general article about TalkSafe and toxic language.