Is Grok AI Safe? UK Guide

If you have seen Grok in the news recently, the stories were not flattering. In January 2026, the AI chatbot made headlines across the UK after it was used to generate non-consensual sexual images of real people. The government got involved, regulators opened investigations, and it turned into a very public and uncomfortable year for Elon Musk's AI company, xAI.

That context matters. It does not mean Grok is off-limits for everyday questions, but it does mean you are going in with eyes open, which is always the right approach. Read this and you will know what to weigh up.

What Is Grok AI?

Grok is an AI chatbot, like ChatGPT or Claude. You type something and it writes back in plain text.

What makes it different is the connection to X, the platform formerly known as Twitter. Grok can pull live information from the web, including news from today, which tools like ChatGPT cannot always do on the free plan. If you already have an X account, you may have found Grok appearing there without looking for it. Free users get a limited number of messages every couple of hours; X Premium and SuperGrok subscribers get considerably more.

There is also a standalone app and website at grok.com. You do not need an X account to use it, though you will need to create an xAI account.

At WellWired we tested Grok for a range of everyday tasks and read through xAI's privacy policy directly. Honestly, it works well enough for factual questions and writing help. But the privacy picture is murkier than ChatGPT or Claude, and the regulatory history we are about to explain gives us genuine pause.

Why Did the UK ICO Investigate Grok?

In the ten days from 29 December 2025, researchers estimated Grok was used to create approximately three million non-consensual intimate images of real people. Some depicted children. UK Prime Minister Sir Keir Starmer addressed Parliament, calling it "disgusting and shameful", and the government announced new legislation to criminalise the creation of such images.

The Information Commissioner's Office, or ICO, is the UK's data protection regulator. It does not usually move quickly. In this case it did. On 3 February 2026, the ICO announced formal investigations into both X Internet Unlimited Company and X.AI LLC, focusing on how they processed personal data when building and deploying Grok. The central question: were proper safeguards designed into the system from the start, or were they an afterthought?

Ofcom also opened its own probe into X under the Online Safety Act. Legal analysis from Simmons & Simmons noted that X faced simultaneous scrutiny under data protection law and online safety law, which is unusual and significant.

Both investigations are ongoing as of July 2026. The ICO has the power to issue fines of up to £17.5 million or 4% of global annual turnover, whichever is higher. According to Hunton's privacy law analysis, the ICO's concern goes beyond the scandal itself. It is asking whether personal data protections were woven into Grok's design, not bolted on after problems emerged.

None of this is about your everyday questions being reported anywhere. It is a question of whether xAI built this responsibly. That distinction matters when you are deciding whether to use it.

Is Grok Safe to Use Now?

X has moved since January. Variety reported that the company deployed controls to prevent Grok from generating sexually explicit imagery, including a block on altering images of real people in revealing ways. Geoblocking is now in place in countries where creating non-consensual intimate images is illegal.

For text conversations (asking questions, getting writing help, or looking things up), none of these restrictions touch you.

The ICO investigation has not concluded. The underlying data questions remain open. That does not make Grok dangerous for everyday use, but you are using a service that UK regulators are actively scrutinising, and that context is worth having before you start.

In our view, Grok is fine for low-stakes questions. Ask it something factual, get help with a birthday message, find out what is in the news. Treat it the same way you would any online service from a company you are still deciding whether to trust, and keep personal information out of the conversation.

What Data Does Grok Collect About You?

According to xAI's privacy policy, the company collects everything you type, every reply it generates, and information tied to your account. If you access Grok through X, it may also draw on your public posts to inform its responses.

By default, xAI can use your conversations to train future versions of the AI. ChatGPT works similarly on the free plan, so this part is not unusual. The main difference is that xAI's data practices are currently under formal ICO investigation, meaning the UK regulator is actively examining whether they meet UK legal standards.

Grok's web app is currently inaccessible in EU countries, following regulatory action over the January 2026 scandal. UK users are not currently blocked, but UK GDPR rules still apply to how your data must be handled, regardless.

If you have an X account, your public posts on that platform count as training data for Grok unless you opt out. Private accounts are excluded by default. If you created an X account years ago and have not logged in since, it is worth checking your settings there before you start using Grok.

What Should You Never Type into Grok?

The rule is the same here as with any AI chatbot. Do not type anything you would not want a company to store on their servers indefinitely.

• Your NHS number or National Insurance number. These identifiers are hard to replace if misused, and no AI tool needs them for anything.
• Bank account details, sort codes, or card numbers. There is never a legitimate reason to share these with a chatbot.
• Passwords or security PINs. Not with Grok, not with ChatGPT, not with anything.
• Passport or driving licence numbers. Same level of caution as financial details.
• Other people's personal information. If you are writing something about a family member, leave out their name and identifying details.

Our guide on what not to share with any AI chatbot covers the full list. Everything in there applies to Grok too.

There is one thing that applies specifically to Grok. Because it is connected to X, anything you type could potentially be associated with your social media presence. If you use X under your real name and post publicly, think carefully about what you discuss in Grok. The two are linked.

If you are worried about scammers using AI tools to target people, our guide on AI scam phone calls in the UK explains the most common tactics and how to protect yourself.

How to Control What Grok Stores

Start with Private Chat mode. You can turn it on in the Grok app or at grok.com. Conversations held in Private Chat are not used for AI training and are deleted from xAI's systems within 30 days. They do not appear in your conversation history either.

To opt out of training altogether, open your xAI account settings and look for the data or privacy section. Once switched off, future conversations will not be used. This does not affect anything already stored.

If you want to stop your X posts being used to train Grok, go into X account settings, find privacy and safety, and look for the AI data option. When you find it, switch it off.

You can delete individual conversations directly inside the Grok interface. For step-by-step guidance on managing privacy in AI tools, see our guide to deleting ChatGPT history; the process is similar and the principles carry across.

Should UK Seniors Consider Using Grok?

Honestly, probably not as a first choice.

Grok's main advantage is live information. If you want today's news or real-time updates, it is one of the few free tools that can genuinely deliver that. For writing help and answering questions it is also capable.

But there are two specific concerns for older UK users. First, the regulatory situation. The ICO investigation means you are using a service the UK's data watchdog is actively scrutinising, which is not something that can be said of ChatGPT or Claude right now. That context should inform the choice.

Second, the X connection complicates things. Many people have X accounts they set up years ago, barely touched since. If that describes you, those old public posts are potentially being used to train the AI you are chatting with. Worth checking before you start.

For most everyday tasks, ChatGPT and Claude are both simpler choices right now. Neither requires an X account, both have clear UK-facing privacy controls, and neither is under active ICO investigation.

If you do prefer Grok, the steps are simple. Turn on Private Chat mode, opt out of training in your settings, and keep personal details out of the chat. You can find more guidance in our AI safety hub.

If you want to feel fully confident using AI tools, the WellWired Academy covers privacy, safety, and everyday use in plain English, with no technical knowledge assumed.

FAQ