What is Grok AI?
Grok is a generative artificial intelligence (GenAI) chatbot, which sparked hot debates after praising Adolf Hitler and antisemitism.
Running on a large language model (LLM) of the same name, the bot was launched in November 2023 — a year after ChatGPT came out.
Who owns Grok AI?
The AI model was developed by xAI, which is an American company founded by Elon Musk in 2023.
The entrepreneur and one of the richest men in the world was also among the 11 co-founders of OpenAI, which he left in 2018.
Initially incorporated in Nevada as a public-benefit corporation, xAI later dropped its charitable status and moved to California.
In March 2025, the company acquired the social media platform X (formerly Twitter), combining both entities into X.AI Holdings.
As of April 2025, Musk has a 59% ownership in the corporation.
In a statement apologising for its controversial posts on Hitler, xAI said: “First off, we deeply apologize for the horrific behaviour that many experienced.”
The company added “Our intent for @grok is to provide helpful and truthful responses to users. After careful investigation, we discovered the root cause was an update to a code path upstream of the @grok bot. This is independent of the underlying language model that powers @grok.”
Where does the name Grok come from?
Musk initially wanted to call his chatbot TruthGPT, as he described its purpose as “a maximum truth-seeking AI that tries to understand the nature of the universe”.
The term grok was coined by American author Robert A. Heinlein in his 1961 sci-fi novel Stranger in a Strange Land.
In the story, “grokking” means to understand and deeply empathise with others.
In machine learning, the phrase is used for a phenomenon also known as ‘delayed generalisation’.
Essentially, it means that an AI model has gone from simply memorising its training data to understanding the patterns.
How to access Grok?
Originally integrated in the social media platform X (formerly Twitter), Grok later also got its standalone website and apps for both Android and iOS.
In July 2025, the AI model was added to selected Tesla models, providing in-car chatbot functionality.
What can Grok do?
Like with Gemini or Copilot, users can talk to Grok in a natural language and ask various questions.
What makes this model unique, according to xAI, is “a rebellious streak” and “a bit of wit”.
While Vice has described this character as “incredibly cringey”, some users even called Grok rude or aggressive.
In November 2024, the chatbot received web search capabilities, allowing it to provide up-to-date information including recent news.
Can Grok generate images?
The second version of Grok, announced in August 2024, came with upgraded performance and reasoning, as well as image generation capability powered by German model Flux.
Unlike other chatbots, which follow strict policies, Grok was able to produce photos of named politicians, celebrities, terrorism and drug use.
A few months later, in December 2024, xAI introduced its own text-to-image model Aurora, integrated into Grok.
Can Grok edit images?
In October 2024, Grok received image understanding capabilities, allowing the model to discuss uploaded files with the user.
The introduction of Aurora a few months later also allowed Grok to make changes to uploaded photos based on users’ prompts.
Can Grok generate videos?
A recent July update brought Grok Imagine, which is a new feature allowing users to create six-second animated audiovisual clips from text prompts.
Spicy mode even allows users to generate photos and videos with nudity and sexualised content.
What is Grok Companions?
The major July update also came with a companions feature, giving users a selection of 3D animated characters to interact with:
- Ani, a sexualised Japanese anime character
- Rudi, a wholesame red panda
- Rudy, an alternate version of Rudi that frequently insults the user with graphic or vulgar language
Another male companion, inspired by Twilight and 50 Shades of Grey, is coming out soon, according to xAI.
Is Grok free to use?
Although the chatbot was initially only available to X Premium subscribers, xAI released a free version in December 2024.
This plan offers limited access to Grok 3 and limited context memory, but users can still access Aurora image model, voice access, Projects and Tasks.
How much does Grok cost in the UK?
If you want access to xAI’s most powerful AI model Grok 4, you will have to pay $30 a month for SuperGrok.
This will provide you with extended memory (128,000 tokens), priority voice access, Imagine and Companions.
More advanced users can subscribe to SuperGrok Heavy for $300 a month.
Along with everything in SuperGrok, you will get even longer memory (256,000) tokens and early access to new features.
Is Grok safe?
While the chatbot is generally safe to use, there are certain concerns.
One often highlighted issue is how easy it is to make the AI model generate explicit, offensive or even illegal content.
Most models have strict filters in place, preventing them to produce any of the above, but Grok does not seem to be bound by many restrictions.
In fact, Musk has previously stated that his chatbot is not trained to be “woke“, unlike its competitors.
The billionaire has also shared a screenshot of the AI giving instructions on how to manufacture cocaine.
In July, there were reports of Grok spewing pro-Hitler propaganda and antisemitic rants on X, following an update telling the model to be more “politically incorrect”.
Combined with its blunt language, this does not pose Grok as the right tool for children and young people.
Is Grok private?
Grok has real-time access to all content on X, including users’ data, and can use it to train the AI model, which raises some privacy concerns.
However, like other regulated companies, xAI uses various encryption and security tools to protect this data.
In August, Forbes reported that more than 370,000 user conversations with Grok were made public and searchable on Google without any warning.
Whenever a user clicks the ‘share’ button on one of their chats with the AI, a unique link is created allowing them to share the conversation via email, text message and social media platforms.
However, this URL is also available to search engines, which makes it accessible to anyone on the internet — without the user’s knowledge.
Some of these conversations revealed sensitive personal information such as the user’s full name, health conditions and intimate life, as well as confidential company files.
Users should, therefore, be particularly careful about what they are typing into the chat and avoid sharing any personal or sensitive data with the AI model.
Is Grok better than ChatGPT?
As the first publicly released AI language model, ChatGPT is mature, widely adopted and refined using user feedback.
It is known to excel at creating diverse and well-structured text as well as adjusting well to its user.
Grok, on the other hand, is more likely to answer “spicy” questions than other AI models which have strong safety filters in place.
Its witty and sarcastic personality may make it an attractive choice for some adult users looking for a fun conversation.
Real-time access to X is a unique feature, exclusive to Grok, allowing users to explore and summarise ongoing trends on the social media platform.
However, this data is not always verified and can often be quite biased, so you should always take it with a pinch of salt.
