Meta's new policy tells parents about their child's use of Meta AI. Only parents who use Instagram supervision controls will get the notification. Later, caregivers using parental controls for other Meta apps will also get alerts.

Meta worked with parents and experts to decide which AI conversations need an alert. This includes conversations where a teen talks about hurting themselves.

Meta AI already gives crisis helplines and encourages teens to seek help. Now, parents will get an alert about these conversations. Meta will be careful when sending alerts.

Artificial intelligence flags concerning chats, and Meta reviews them before sending an alert. The parent also gets suggestions for discussing self-harm and suicide with their child.

Parents will get the alert via an app notification and a separate email, text, or WhatsApp message.

In February, Meta launched a similar notification feature for Instagram.

A doctor from the Center for Suicide Prevention and Research welcomed the new protections for teens. He said the feature could be helpful, but only if notifications are easy to use and lead to change.

Meta will send alerts to parents in the US, UK, Canada, and Australia. It will be available everywhere by the end of the year.

A staff attorney for the children's advocacy nonprofit Fairplay said Meta's announcement is a step in the right direction, but should be viewed with skepticism.

The attorney said Fairplay's research found that Meta's safety tools are not effective. Meta said the report was incorrect.

Meta also announced a limited content setting for Meta AI. This feature debuted for Instagram last year.

The limited content setting enables Meta's strictest filters. Meta AI is already trained not to engage in certain conversations with teens.

When the limited content setting is turned on, Meta's models become more aggressive in identifying problematic prompts. They will decline to answer a broader range of queries.

Child safety advocates have criticized Meta's recent parental control updates as not enough.

In April, Meta gave parents some insight into their child's conversations with Meta AI. The feature highlights broader topics, such as school and health.

A Fairplay executive director said the feature burdens caregivers with monitoring their child's online activity instead of making a safe product.

Meta continues to face legal scrutiny for its performance on child safety. The company lost two landmark trials and will appeal both verdicts.

This story has been updated with additional information from Meta and Fairplay.

If you're feeling suicidal, please talk to somebody. You can call or text the 988 Suicide & Crisis Lifeline or chat at 988lifeline.org. There are also other resources available, such as the Trans Lifeline and the Trevor Project.

Topics include Mental Health, Social Good, Family & Parenting, and Meta.