Internet safety for kids and teens, updated for AI

1. Keep personal details private.

Personal information covers more than a name. A US privacy rule for children counts a full name, address, phone number, email address, location, photos, videos and voice recordings.

• Ages 8 to 10: Your name, address, school and photos stay off the internet unless a grown-up says yes.
• Ages 11 to 13: Turn off location sharing in apps that don't need it, and check who can see your profile.
• Ages 14 to 18: Small details add up. A school logo, a street sign and a username can be enough for a stranger to work out where you are.

2. Use strong passwords.

A US cybersecurity agency gives three rules: make passwords long (at least 16 characters), random and different for every account.

• Ages 8 to 10: A grown-up keeps your passwords. Don't share them with friends, even best friends.
• Ages 11 to 13: Use a long, random password for each account, and never reuse one.
• Ages 14 to 18: Use a password manager, a program that makes and stores strong passwords so you remember only one.

3. Tell a trusted adult.

This is the tip that makes the others work.

• Ages 8 to 10: If something online makes you feel scared, sad or confused, stop and tell a grown-up. You won't be in trouble for telling.
• Ages 11 to 13: Pick two adults you would tell, one at home and one outside it, such as a teacher or a relative.
• Ages 14 to 18: Telling someone is not giving up control. An adult can help you report, block and keep evidence.

4. Think before you post.

Once something is shared, someone may already have copied it, even if you delete it.

• Ages 8 to 10: Ask a grown-up before you post or send any picture.
• Ages 11 to 13: Ask yourself: would I be fine if my whole class saw this?
• Ages 14 to 18: Ask before posting a picture of someone else, and expect the same from them.

5. Be kind, and don't pass on cruelty.

A US government guide on cyberbullying says: don't respond to or forward bullying messages, keep evidence, and block the person.

• Ages 8 to 10: If someone is mean online, don't reply. Show a grown-up.
• Ages 11 to 13: Take a screenshot before you block, so there is a record.
• Ages 14 to 18: Most apps have a report button, because bullying often breaks their rules.

6. Remember that people online may not be who they say.

A profile picture and a friendly message prove nothing about who is typing.

• Ages 8 to 10: Only chat with people you know in real life, in apps a grown-up set up.
• Ages 11 to 13: If someone asks you to keep a secret, move to a private chat, or send a picture, stop and tell an adult.
• Ages 14 to 18: Pressure, flattery that moves fast, or threats are signals to stop and get help, not to handle it alone.

7. AI answers can be made up.

A US government guide on AI risks describes AI systems producing "confidently stated but erroneous or false content". It also notes they can make up reasons or sources to back an answer.

• Ages 8 to 10: AI can sound sure and still be wrong. Check with a book, a teacher or a grown-up.
• Ages 11 to 13: Check anything that matters with a second source that a person wrote.
• Ages 14 to 18: If an AI answer cites a source, open the source. Sometimes it doesn't exist, or it says something else.

8. Voices and pictures can be faked.

AI can make realistic fake videos, pictures and voices, often called deepfakes. The US consumer protection agency warns that a scammer may need only a short audio clip to copy a family member's voice, and says: don't trust the voice; call the person back on a number you know is theirs.

• Ages 8 to 10: If a video or voice message seems strange, show a grown-up before you believe it or share it.
• Ages 11 to 13: Before believing a shocking picture, look for the same story from a news source you trust.
• Ages 14 to 18: Never make or share a fake picture, video or voice of a real person. It can hurt them, even as a joke.

9. Share less with AI apps.

Apps may ask for your location, contacts or photos. Some AI chat apps save chats, and one maker's help page says some chats are read by human reviewers.

• Ages 8 to 10: Never type your name, address, school or worries into a chatbot, and never upload a photo of your face.
• Ages 11 to 13: With a grown-up, check what each app is allowed to use, and turn off what it doesn't need.
• Ages 14 to 18: Find the history and training settings in any AI app you use, and decide what you want saved.

10. Watch for hidden instructions.

Some AI tools read web pages, files or emails for you. A security standards group explains that text inside those sources can change what the AI does, which is called indirect prompt injection. One of its example cases is a web page with hidden instructions that change what an AI tool does when asked to summarize it.

• Ages 8 to 10: Use AI tools only with a grown-up nearby.
• Ages 11 to 13: If an AI summary of a page says something odd, the page itself may be steering it. Read the page.
• Ages 14 to 18: Don't let an AI tool send messages, buy things or share files without checking each step yourself. The same group's advice for builders is to require a person's approval for high-risk actions.

They cover:

• The everyday rules for ages 8 to 18, with the four AI risks families meet most often in the news: made-up answers, fakes, data sharing and hidden instructions.
• Where to report, using the official services' own pages.

They don't cover:

• Every app's settings. For chatbot settings and parental controls, see the checklist guide linked below.
• Legal advice, or what to do in a crisis beyond the first steps. In an emergency, call 911.