Is Zeta AI safe? Age rules, filters, chat privacy and Korea's hearing

Is Zeta safe for kids and teens?

Not for young children, and only with supervision for teens. Zeta’s terms bar users under 13 and require a parent or guardian to agree on behalf of anyone aged 13 to 17. Its help center says the platform is meant for all ages, including minors, and that it strictly removes sexual or violent plots. Outside reporting says the filter has gaps. In September 2025 a Scatter Lab lawyer told The Korea Herald its filters are not perfect, and in October 2026 Korean lawmakers said unverified accounts still received self-harm and sexual content (see below).

Use is heavy. The Korea Herald reported that Korean users averaged 2 hours 46 minutes a day in June 2025, and that most users are teens or in their 20s. Zeta has no parent dashboard or linked parent account, and its terms make parents responsible for supervision.

What age is Zeta for?

Zeta’s own minimum age is 13, but the store ratings disagree with each other. As listed on October 19, 2026:

RoboRhythms reports that Apple’s rating dropped from 17+ to 13+ with an August 14, 2026 update that also tightened the filter. Zeta hasn’t published an official note on that change.

How does Zeta’s filter work?

Zeta filters every conversation with AI and defaults to Safe Mode. Its terms ban content sexualizing minors, non-consensual imagery, hate, self-harm promotion and, outside one exception, sexually explicit material. That exception is an adults-only “Unlimited Mode,” which the app lists for users 18 and over and the terms say must not be reachable by anyone known or believed to be under 18.

In Korea, unlocking it takes a phone-based identity check that lasts one year. Zeta’s Korean help center says overseas residents can’t complete that check yet, and its English help center says it has declined to build an 18+ service. Couldn’t confirm whether or how US users can turn Unlimited Mode on. Users can report plots and individual replies in the app, and the terms point anyone at risk to the 988 crisis line in the US and Canada.

Does Zeta read your chats?

Zeta’s systems read every chat; its staff say they don’t, as a rule. The help center says conversations aren’t shared with other users or plot creators unless you publish them and that the operations team doesn’t view chats individually, but mandatory AI monitoring checks every conversation against its rules. The privacy policy adds that Zeta records and stores conversations, uses them to train its AI and may hand data to authorities when the law requires.

What does Zeta do with your data?

Zeta collects a typical app’s worth of data and uses chats for AI training by default. What its privacy policy, vendor list and store labels say:

Ads are their own risk. Zeta’s help center says ads come from outside networks and some unsuitable ones get past its filter, and RoboRhythms reports fake warning ads on the free tier.

What happened at Korea’s October 2026 hearing?

Korean lawmakers criticized Zeta’s youth protections at a National Assembly audit on October 12, 2026, as reported by DigitalToday. Their allegations:

• One office said an account without adult verification, with safe mode on, still produced conversations describing self-harm and suicide.
• Another said sexual scenes involving minor characters were generated without adult verification, and the AI explained how to get around blocking.
• A lawmaker said Zeta logs about 110 million hours of use a month in Korea, more than twice ChatGPT, and called for age checks and time limits.

Scatter Lab reportedly began strengthening its blocking. ZDNet Korea reports the company’s plan to cap teens’ basic daily use at 3 hours and their monthly spending at 100,000 won; we couldn’t confirm it on Zeta’s own pages. CEO Kim Jong-yoon missed the October 12 audit for a US trip.

He was then called as a witness at the Education Committee’s October 23, 2026 audit on youth chatbot issues, but filed a written excuse citing a business trip to Japan and did not appear, ZDNet Korea and Newspim report. At that session a lawmaker said his excuse lacked supporting evidence, and the committee chair said absent witnesses’ excuses would be reviewed and unsupported ones referred for a formal complaint. As of October 24, 2026, no decision on a complaint had been reported.

What was Lee Luda, and does it matter?

Lee Luda was Scatter Lab’s earlier chatbot, and it led to Korea’s first privacy sanction against an AI company. On April 28, 2021, the Personal Information Protection Commission fined Scatter Lab a total of 103.3 million won. It found that about 9.4 billion KakaoTalk sentences from roughly 600,000 users of the company’s relationship-analysis apps were used to build the bot without proper consent or anonymization, and that children’s data was collected without guardian consent. The bot was shut down in under three weeks. Scatter Lab told The Korea Herald it has since built new anonymization processes. Zeta’s current privacy policy at least discloses that chats train its AI and offers an opt-out by email.

How can parents and users stay safer on Zeta?

Lock down the device first, because Zeta has no parent controls of its own. Steps that help:

1. Set age limits and download approval in Apple Screen Time or Google Family Link.
2. Block zeta-ai.io in the browser or router if you block the app, since the web version works too.
3. Email privacy@scatterlab.co.kr with “AI Training Opt-Out” to keep future chats out of training.
4. Limit ad tracking in your phone’s privacy settings, and don’t tap buttons inside suspicious ads.
5. Keep real names, addresses and photos out of chats and public plots.
6. Talk about spending, since Pieces and the $14.99 zeta pass are a tap away.

For how Zeta compares overall, read our Zeta AI review and Zeta vs Character.AI, or see Is Character.AI safe? and Is Talkie AI safe?.