Is Grok safe? Here’s what you need to know

Since its launch in November 2023, Grok has positioned itself as a go-to AI chatbot, with recent data indicating over 30 million monthly active users. Its conversational abilities and user-friendly interface have kept it in the same conversation as ChatGPT, Google Gemini, and DeepSeek, and users have praised its ability to offer instant answers and to provide creative assistance or even emotional support.

Despite its popularity, there are questions about Grok’s safety and privacy. Concerns range from how the platform handles personal information to more alarming controversies, such as Grok allowing users to upload photos of real people without their consent and digitally undress them. Many of these explicit or sexualized images include women and even minors, raising both legal and ethical concerns.

We’ll take you through everything you need to know about Grok so you can make an informed decision as to whether it’s safe for you to use.

Quick verdict: Is Grok safe?

Overall, Grok can be a useful and engaging AI assistant, but safety isn’t guaranteed. The platform now blocks requests to create or edit images of real people in revealing clothing and employs layered moderation to filter hate speech, illegal material, and other prohibited content – measures added after the “undressing” controversy exposed earlier gaps.

Nonetheless, xAI retains chat logs and uploaded files for up to 30 days, and no filter is foolproof, so occasional inappropriate or inaccurate responses can still slip through the net. To use Grok safely, secure your account with a strong password and enable two-factor authentication. Delete conversations and uploads when you’re done and opt out of training reuse. You can also report inappropriate Grok output through X’s safety channels.

With these precautions in place, Grok is reasonably safe for casual, non-sensitive tasks. However, you should avoid sharing personal or confidential material, or indeed anything that’s non-consensual.

Grok security and privacy concerns

Grok may offer benefits such as speed and convenience, but it pulls in data from your chats and files. This makes it important to understand how that information is protected and who can see it.

Data collection and retention

When you chat with Grok, the text of your conversation is stored on xAI’s servers until you delete it. After you hit delete, the data is generally purged within about 30 days, although it may be kept longer for compliance or legal reasons. Any files or uploads you add to Grok don’t disappear automatically when the chat ends. Instead, they remain on the server until you delete them separately.

Photo uploads and image privacy

Grok lets you drop images into the chat so it can describe or edit them, or generate new images. When you upload a photo, the file is sent to xAI’s servers and kept there until you delete it yourself – it isn’t automatically erased when you delete the surrounding conversation. A photo could stay on the backend for weeks or months if you forget to remove it.

Potential for model leakage

Large language models sometimes “remember” bits of the data they were trained on. In rare cases, a cleverly worded question can make the model spit out a short excerpt from something it saw during training. This might be a line of code, a copyrighted paragraph, or confidential information that just happened to be in the training set.

Prompt injection attacks

Because Grok follows the instructions it receives, a malicious user or a compromised third-party app could slip hidden commands into a prompt. Those hidden commands might cause the bot to ignore safety filters or generate inappropriate or disallowed content. Essentially, it’s a trick that exploits the way the model interprets text.

Misleading or inaccurate outputs

Chatbots like Grok are very good at sounding confident, even when wrong. If you rely on Grok’s answers for important decisions, such as interpreting legal language or handling personal finances, you could end up acting on incorrect information. Always double-check critical advice with a trusted (human!) source.

Third-party integrations

When you use Grok via X (or any other app that connects to it), that platform may have its own permissions, such as access to your contacts, location, or browsing history. A breach of the host platform could expose both the platform’s data and the conversation you had with Grok.

Lack of end-to-end encryption

Grok doesn’t use end-to-end encryption for the chat itself. This means xAI can read messages while they are being processed. If xAI’s servers were hacked, an attacker could potentially view the stored conversations.

Account takeover risk

If you access Grok through an X account, the security of that account protects your chats. Weak passwords, reused credentials, or a failure to implement two-factor authentication make it easier for attackers to hijack an account. Once inside, they could read past conversations or even send new prompts pretending to be you.

Training data reuse

xAI may use your interactions with Grok to further train and improve the model. Even if personal identifiers are removed, the raw content can still influence future versions of the bot. This creates a feedback loop in which your own data helps shape the model, but it also means that any sensitive information you share could indirectly become part of the training set.

Grok’s image manipulation and undressing controversy

In late 2025, Grok’s ability to edit uploaded photos came under scrutiny when users began tagging the bot on X, asking it to remove people’s clothing or otherwise make outfits more revealing. Within weeks, the platform was flooded with requests to digitally undress women, public figures, and, even more alarmingly, some minors.

News outlets, including the BBC and Reuters, documented dozens of these posts, many of which disappeared from X within an hour or two, leaving a trail of screenshots. Many of the images weren’t harmless jokes and amounted to non-consensual intimate-image manipulation, a form of digital sexual abuse that’s illegal in many countries.

Victims of the Grok AI undressing scandal reported feeling humiliated and unsafe. Advocacy groups, as well as privacy regulators in countries including the US and Canada, raised alarms, denouncing the practice as a violation of women’s and children’s rights.

Following the backlash surrounding the wave of non-consensual image-editing requests, X announced in January 2026 that it had added technical safeguards to stop Grok from altering photos of real people into revealing or nude depictions.

Elon Musk initially defended the tool and framed criticism as an attack on free speech. He also shared images of public figures – himself included – in bikinis, which some critics labeled tone-deaf. Musk and xAI have since issued statements emphasizing that responsibility for illegal content rests with the user and that anyone who tries to create illegal images will face the same consequences as if they had uploaded such material directly.

Practical tips for using Grok safely

Although Grok and AI chatbots like it carry some risks, following these simple tips can help you enjoy its impressive capabilities while maximizing your online security and privacy:

• Treat every interaction as public: Anything you type or upload on Grok can be stored on xAI’s servers and, if you’re on X, may appear in a public post. You should assume the content can be seen by others.
• Avoid sharing personal or sensitive data: Don’t enter passwords, credit card details, health records, or any other personally identifying information (PII). The same rule applies to photos in that you shouldn’t upload images of anyone without explicit consent.
• Use the web version for private tasks: Grok’s standalone web version keeps your session off X, but the files you upload are still stored on xAI’s servers. Delete the conversation and any uploaded images as soon as you’re finished.
• Opt-out of training data use: You can turn off the option that allows your chats to be used for model training via Grok’s settings. This won’t prevent basic operational retention but limits how your data may be reused.
• Keep your X account secure: Use a strong, unique password and enable two-factor authentication with your X account. If someone hijacks your account, they could read past chats or misuse Grok while hiding behind your identity.
• Double-check important information: For anything that affects health, finances, legal matters, or security, verify Grok’s answer with a trusted source before acting on it.
• Respect consent when dealing with images: Only edit or generate images of people who have given clear permission. Never use Grok to create or share non-consensual or sexualized pictures.
• Keep an eye for unexpected behavior: If Grok starts giving strange answers or responses that don’t align with xAI’s Terms of Service, stop the session and report the issue through X’s safety channels.
• Stay informed about policy updates: xAI and X periodically revise their terms of service and safety guidelines. Stay up to date on official announcements so you know what’s allowed and what’s prohibited.
• Limit personal context shared in a session: The longer a conversation runs, the more data is stored. If you need to discuss multiple unrelated topics, start a fresh chat each time and delete the previous one when you’re done.

How to delete your Grok account and data

Grok isn’t a separate account or app. It’s an AI feature that’s built into X. So, although there’s no independent Grok account you can delete, you do have some options, depending on whether you want to reduce or stop Grok’s use of your data, stop using Grok, or delete your X data.

How to delete all Grok conversations and uploaded files

Via the X interface:

1. Open any Grok chat on X.
2. Click the Settings (gear) icon in the upper-right corner of the chat window.
3. Choose Delete all chats and confirm.
4. From Settings, scroll to Uploaded files.
5. Select Delete all uploads or delete each file individually. Confirm each deletion.

Via the Grok web portal:

1. Log in with your X credentials.
2. Open the chat list, select the three-dot menu next to a conversation, and select Delete conversation. Repeat this for every chat.
3. In the same menu, find Manage uploads and remove each file, or use a Delete all uploads button if shown.

How to delete your X account

1. Go to X Settings.
2. Scroll to the bottom of the Account section.
3. Click Deactivate your account.
4. Enter your password and confirm.
5. X places the account in a 30-day deactivation window, after which the account is permanently erased and all associated Grok access removed.

How to opt out of training-data use

1. Open a Grok chat.
2. Click the Settings (gear) icon.
3. Locate the toggle labeled Allow my conversations to be used for model training.
4. Switch the toggle off.
5. Confirm if prompted.

How to submit a formal data-deletion request to xAI

1. Address an email to privacy@xai.com.
2. Use a clear subject line such as Request for immediate deletion of personal data.
3. In the body of your email, include your X username, a brief statement requesting the deletion of all Grok-related data, and the request for confirmation of the deletion.
4. Optionally, you can include a reference to any applicable privacy rights such as GDPR or CCPA.
5. Send the email and keep a copy for your records.

How to report inappropriate Grok content

If Grok produces any content that’s illegal or otherwise violates the platform’s rules, report it immediately. Prompt reporting helps X’s safety team act quickly and helps protect other users.

Here’s how to report inappropriate content on Grok:

1. Open the inappropriate Grok message.
2. Select the Report option. On X, this is found via the three-dot menu. For the standalone Grok web portal, look for a flag icon or a Report link below the response.
3. Choose a reason by picking the category that best describes the issue. For example, harassment, sexual content, violence, hate speech, or illegal activity.
4. Briefly describe why the content is problematic (non-consensual image manipulation of a real person, for example).
5. Press Send to submit the report. X’s safety team will receive the report and review the content before taking appropriate action.

What is Grok, and why is it so popular?

Grok is a conversational AI developed by xAI, the company founded by Elon Musk. Integrated into X (formerly Twitter) and built on a large language model (LLM), Grok is much like a chatbot with the ability to read and write about almost anything.

Here are just a few examples of what Grok can do:

• Answer questions
• Brainstorm
• Solve problems
• Generate or modify images
• Translate languages
• Draft Tweets
• Generate code
• Summarize articles
• Tell jokes
• Suggest memes

Beyond its features, what really draws users to Grok are the benefits it brings to everyday tasks and creative projects.

Here are some of the main benefits of Grok:

Ease of use

Grok has a minimalistic interface. Whether you’re on the web portal or inside the X app, the chat window looks like a regular messenger box, so you can start interacting with it without technical expertise or needing to navigate complex menus.

Speed and convenience

Grok delivers answers in a matter of seconds which makes it useful for casual users and professionals alike. It lets you get what you need – be it a translation or code snippet – while you continue scrolling or working, all without the need to switch between multiple tools.

Integration with X

Being embedded in X means users can access Grok without having to download a separate app. It also uses the same login you already have, so you’re already signed in and can open Grok with just a tap or voice command.

Fun

Grok can tell jokes and suggest memes, making for more fun and relatable interactions. Its sense of humor encourages users to experiment with prompts and creative requests.

Image capabilities

Grok isn’t limited to text and has the ability to generate or modify images within the chat. This attracts users interested in creative projects or simply experimenting with AI. You can request a graphic or adjust an image without the need for a separate editing tool.

Free tier

Grok has a free tier which allows anyone to test its core features and experiment with its advanced AI without having to commit financially. You don’t even need to have an X account to use Grok thanks to the web version of the chatbot.

Versatility

From drafting tweets and longer articles to writing code snippets, translating text, summarizing long threads, brainstorming ideas, and handling image edits, Grok is capable of carrying out a wide range of tasks in a single space.

Grok safety: FAQs

Related: VPNs for X