AI Safety: Everything You Need to Know

Understand the technology you’re using, the information you’re sharing and the choices you have. Our AI safety guides explore privacy, security, data, children and the wider risks of everyday AI use, with practical advice grounded in how these tools actually work.

AI Safety: How to Use AI More Safely

AI is becoming part of everyday life incredibly quickly. We use it to search, write, plan, learn, work and increasingly to talk through things we might once have kept to ourselves. That makes understanding what happens behind the chat box more important than ever.At Digital Safety Squad, we look at AI safety from both the technical and human sides. That means examining what services such as ChatGPT, Claude, Gemini and Perplexity collect, how conversations and uploaded files are handled, which privacy and security controls actually make a difference, and where the risks change when children and teenagers are involved.Our guides are built around the questions people are already asking as AI becomes more personal: whether AI chats are private, what you should never share with a chatbot, how to delete your data, which services offer stronger privacy protections and what ‘safe’ really means when a tool is learning more about you over time.The aim isn’t to make AI feel frightening or to tell you not to use it. It’s to give you enough information to understand the trade-offs, make deliberate choices about what you share and use these tools without giving away more of yourself than you intended.

On this page

How AI Changed What Safety Means Online

A few years ago, digital safety mostly meant strong passwords and spotting a scam email. Now it also means knowing what happens after you type something into ChatGPT, whether Gemini can see your Gmail, and whether your teenager's late-night conversation with an AI chatbot is something to worry about.

AI chatbots have moved from novelty to daily habit faster than almost any technology before them, and most people are using several without ever comparing how differently each one handles their data.

AI Use in 2026, at a Glance

More than half of UK children aged 8 to 17 now use AI tools regularly, according to Internet Matters.

Around two-thirds of UK children aged 9 to 17 use AI chatbots regularly, with vulnerable children more likely to use them in a companion-like way.

ChatGPT is now one of the most impersonated brands in phishing attempts, according to Check Point's Q2 2026 Brand Phishing Report.

Most people have never compared how ChatGPT, Claude, Gemini and Perplexity actually differ on privacy, despite using more than one of them.

This hub is organised around the actual questions people bring to us: which tool to trust, what to keep private, how to spot an AI-enabled scam, what this means for your kids, what it means at work, and what it's doing to relationships and mental health.

As AI tools keep changing, we'll keep this hub updated with what we find when we actually check.

Which AI Tool Are You Using?

Start here if you already know which tool you want to check. Each guide covers what it collects, how secure it is, what happens to your conversations, and what we'd actually change first.

Tool Safety

Is ChatGPT Safe?

The most widely used AI tool

Privacy, security, uploads, account tiers and the settings worth changing first, checked directly against the product.

Read the full guide
Tool Safety

Is Claude Safe?

The strongest training-consent default

What allowing or declining model training actually changes, and why Claude's consumer service is restricted to adults.

Read the full guide
Tool Safety

Is Google Gemini Safe?

The deepest Google integration

Keep Activity, Connected Apps, and what being tied to your wider Google account actually means for privacy.

Read the full guide
Tool Safety

Is Perplexity Safe?

Search-first, different risks

Perplexity Search, the mobile app and the Comet browser carry genuinely different risks worth knowing apart.

Read the full guide

Protecting Your Privacy Across Any AI Tool

These questions come up no matter which chatbot you're actually using. This section pulls together our guides on what to keep out of any AI chat, what 'private' actually means, and what to do once you've already shared too much.

Prevention

What Information Should You Never Share With AI?

The habit that matters most

A practical share, redact or never-share framework that works across every AI provider, not just one.

Read the full guide
Understanding privacy

Are AI Chats Private?

What 'private' actually means

A proper like-for-like comparison across ChatGPT, Claude, Gemini and Perplexity, not just a single word answer.

Read the full guide
Taking action

How to Delete Your Data From AI Chatbots

What to actually delete, and how

What you actually want removed, provider by provider, plus your UK GDPR right to erasure.

Read the full guide
Recommendations

Best Private AI Chatbots

Hands-on tested, not just claimed

Proton Lumo, DuckDuckGo AI Chat, Brave Leo and mainstream tools configured more privately, tested directly.

Read the full guide

Is AI Safe for Your Family?

It depends heavily on age and which tool, which is exactly why family AI safety deserves proper space rather than a couple of lines tucked into a general guide.

Family safety

Is ChatGPT Safe for Kids?

Our age-based verdict

What parents can and can't control, warning signs of unhealthy use, and what to do if ChatGPT gives your child a harmful response.

Read the full guide
Family safety

AI Chatbots: The Hidden Dangers You Need to Know

The wider picture, beyond one tool

How AI chatbots work in general, and the risks that apply regardless of which one your child is using.

Read the full guide
Companion apps

From Ani to Baby Grok: Are AI Chatbots Safe for Children?

Beyond ChatGPT and Gemini

What AI companion characters like Ani and Baby Grok actually are, and what to know before a younger child starts talking to one.

Read the full guide
Companion apps

AI Girlfriends: The Newest Online Risk for Kids

Why these apps need their own rules

Why AI girlfriend and companion apps carry different, often higher, emotional-attachment risks than a general-purpose chatbot.

Read the full guide
Family safety

Is Your Child Talking to Chatbots? Here's What to Do

Spotting it, and what to say

How to recognise when a child is treating an AI chatbot like a friend, and how to open the conversation without shutting it down.

Read the full guide
Family safety

How to Talk to Your Kids About AI Friends

A script for the conversation

Practical ways to start the conversation about AI companions before it becomes a difficult one.

Read the full guide
Family safety

Age-Appropriate AI Rules

A framework by age group

A practical, age-by-age framework for deciding how much AI independence makes sense for your child.

Read the full guide
Education

AI in Schools: How Parents Can Support

What's actually happening in the classroom

What AI looks like in schools now, and how to support your child through it without falling behind on what's changed.

Read the full guide
Education

How to Spot AI-Generated Homework

Study aid vs doing the work

The difference between AI helping your child understand something and AI doing the work for them, and how to tell which is happening.

Read the full guide
Family safety

What to Do If Your Child Is Bullied With AI

A practical response, not panic

Practical steps if AI has been used to bully, impersonate or humiliate your child online.

Read the full guide

Want the wider family picture beyond ChatGPT?

Our Family Guide to AI covers AI chatbots, companion apps and age-appropriate rules across the whole family, not just one tool.

Read the Family Guide to AI

Is AI Safe to Use at Work?

Yes, with the right guardrails in place, but the risks are different to using AI at home. AI tools bring different risks into a business than they do at home, from what an employee might paste into a chatbot to how convincingly AI can now impersonate a colleague or supplier.

Business AI

What Is AI Safety for Organisations?

Where to actually start

What businesses need to think about before rolling AI tools out to a team, without needing a full IT department to do it.

Read the full guide
Business AI

How to Write an AI Safety Policy

Getting the rules in writing

A practical starting point for putting your organisation's AI rules in writing, rather than leaving it to individual judgement.

Read the full guide
Verification

AI Impersonation: How to Verify Clients & Suppliers

When a familiar voice isn't who it seems

Practical verification steps for a business world where AI can now convincingly imitate a client, supplier or colleague.

Read the full guide
Training

How to Train Employees to Avoid Phishing and Deepfake Scams

Building the habit across a team

How to actually train a team to catch AI-enabled phishing and deepfake attempts, not just send round a policy document.

Read the full guide

Is AI Changing Your Relationships?

Yes, often before either partner has consciously decided how they feel about it. AI isn't only reshaping how we work and parent, it's showing up inside relationships and mental health too, in ways that are easy to miss until they've already changed something.

Digital relationships

AI Relationship Boundaries: How AI Creeps In Without You Noticing

Starting the conversation

Step-by-step prompts for talking about AI companions and chatbot intimacy with a partner, calmly and before it becomes a fight.

Read the full guide
Digital relationships

AI Relationship Red Flags: The Secret Lover You Didn't See Coming

Early signs worth noticing

Behavioural patterns that suggest AI use has shifted from harmless distraction into a hidden emotional escape.

Read the full guide
Digital relationships

Is Emotional Bonding with an AI Partner Cheating?

Setting a line that works for both of you

Questions and frameworks for deciding together what counts as cheating when AI is involved.

Read the full guide
Wellbeing

Is AI Impacting Your Mental Health?

A free quiz and practical resources

How AI chatbots, algorithms and filters shape what we think and feel online, and where a healthy line sits.

Read the full guide

Frequently Asked Questions

Which AI chatbot is safest overall?

There isn't one universal answer, since 'safest' depends on what you're weighing. Claude currently gives consumers the strongest training-consent starting point. Gemini is the most transparent about its data practices. ChatGPT has the broadest safety and account-security tooling. See Are AI Chats Private? for the full comparison.

Is it safe to use AI chatbots for everyday tasks?

Yes, for ordinary writing, planning, brainstorming and research. The caution applies to what you share, not whether you use the tool at all.

Do I need to avoid AI to stay safe online?

No. Avoiding AI entirely isn't necessary or realistic for most people. Understanding what a specific tool does with your data, and being deliberate about what you type into it, achieves far more than avoidance.

Is AI safe for children to use?

It depends heavily on age and which tool. Claude is restricted to users 18 and over. ChatGPT and Gemini allow teenage use with protections in place. See Is ChatGPT Safe for Kids? for the full age-based breakdown.

How often is this AI safety guidance updated?

We recheck settings and factual claims whenever a guide is substantially updated, and each page states when it was last checked, since AI providers change their privacy practices more often than most software.

Our Family Safety Hub has all the information you need to protect your loved. From trusted guides, apps, tools, reviews and expert advice, we break down everything you need to know with simple, practical steps that actually work. Whether you're just getting started or looking to level up your protection, you’ll find clear, friendly guidance every step of the way.