STOP AI SPYING ON YOU

AI tools are endlessly useful but they also collect and analyse your personal information. Robert Irvine reveals every setting you must change to stay private

Generative AI tools such as ChatGPT, Gemini and Copilot have become impossible to avoid and difficult to resist. More than 20 million people in the UK now regularly use them to get information and advice; create text and image content; and summarise long documents and videos (see www.snipca.com/57836).

However, few of us pause to consider that the more we use AI-powered tools, the more information they collect about us. Their simplicity and efficiency may yield fast and (usually) impressive results, but there’s a lack of transparency about what these tools do with the data we

We’re now accustomed to the privacy threats posed by ads and hidden trackers in websites and apps, and are familiar with the methods you can use to combat them. But with new technology comes new concerns, and the rapid rate at which AI is evolving makes it essential to know how to protect your personal data.

In this feature, we explain how to lock down your privacy in popular AI tools to prevent them from collecting more data than they need. Rather than use them without signing in, which limits their features and the quality of their responses, we reveal the settings you must change and the data you should delete, so you can continue to enjoy AI’s benefits without being spied on.

WHAT YOU CAN DO

• Stop chatbots using your data to train their AI models

• Prevent AI tools from remembering your personal details

• Switch to a temporary chat to interact with AI privately

• Delete information that AI tools have already saved about you

• Don’t let AI use your data to show you personalised ads

• Disable intrusive new AI features in your web browser casually provide.

STOP AI CHATBOTS SPYING ON YOU

Interacting with an AI chatbot feels private because it’s like having a one-to-one conversation with a knowledgeable friend. There’s no risk of strangers eavesdropping on your conversations, as there is on social media, and your queries don’t result in you being tracked around the web, like they do when you search with Google.

But nearly all AI assistants collect personal information about you, through the prompts you enter, the files you upload and the commands you speak (if you use voice mode). They also save your chats to train their large language models (LLMs) and deliver more relevant responses.

More worrying are the personal details chatbots infer from your interactions, without you explicitly providing that data. Their sophisticated and constantly evolving LLMs allow them to ‘read between the lines’ in ways far beyond the traditional profile-building associated with Google, Meta and Amazon.

They can deduce your interests, opinions and personality, as well as your location, job, relationships, skill levels and much more. AI assistants don’t just watch what you click and tap. They also learn how you think.

Here, we explain how to limit data collection in the most popular chatbots, so you can enjoy their benefits without them invading your privacy.

ChatGPT

In accordance with GDPR rules, OpenAI’s privacy policy for UK and EU users (www.snipca. com/57765) explains how and why ChatGPT collects your personal data. It states that “when you use our services, we collect personal information that is included in the input, file uploads or feedback that you provide”, and says this data is used to improve its services and train its models. But it doesn’t elaborate on precisely which details it extracts from your chats.

Additionally, ChatGPT users on its free tier and new Go plan will soon start seeing ads (www.snipca.com/57766). These will appear at the bottom of answers “when there’s a relevant sponsored product or service based on your current conversation”.

Although ads will initially be limited to the US, it’s only a matter of time before UK users see them too. OpenAI says your data and conversations are “never sold to advertisers”, but the ads will still be personalised based on your chats.

Tweak the following settings to stay as private as possible when using ChatGPT.

Adjust ChatGPT’s data controls

To stop ChatGPT using the content of your chats to train its AI models, click your profile name in the bottom-left corner of the ChatGPT website and choose Settings. Click ‘Data controls’ then ‘Improve the model for everyone’ ( 1 in our screenshot below) and switch the option off.

The change applies across all ChatGPT apps and will disable the settings that let ChatGPT use your audio and voice recordings for its model-training. To stop it storing your past conversations, click ‘Delete all’ next to ‘Delete all chats’ 2 .