First MechaHitler, Now an Adults-Only Avatar: How Elon Musk Is Building His AI Brand

• Elon Musk is using controversy to encourage the use of his artificial intelligence system, Grok. First, he allowed it to insult users of Portal X. Now he's introducing an erotic avatar into conversations.
• The billionaire entered the language model market with a delay - he founded the xAI company only in March 2023. He was previously an investor in OpenAI, but withdrew from the project.
• Musk, through his companies, is building an entire ecosystem powered by Grok. Tesla cars, the Starlink system, and Optimus humanoid robots will all use it.
• xAI uses supercomputers based on, among others, 200,000 NVIDIA H100 GPUs to train its models and builds its own independent technological infrastructure.

Elon Musk's xAI artificial intelligence is making waves again. This time, the Grok-powered app has launched interactive companions with which users can chat in real time. The new feature is available as part of a paid SuperGrok subscription, which costs $30 per month.

What's the deal with AI companions available in Grok?

Starting Monday, Grok app users can chat with one of two virtual friends. Bad Rudy is a cartoonish red panda programmed by the developers to play tough and encourage people to commit crimes, for example. The second is Ani, a stylized character from Japanese animation. Interactions are conversational, with elements of personalization. The level of the relationship with the avatar influences the available responses and reactions of the character.

The second character in particular is controversial. She was designed to engage in romantic conversations with users. According to those who have already tested the app, a well-paced conversation with Ani can result in the character taking off her clothes and engaging in erotic narration .

Elon Musk's move is a new development in the market of widely available chatbots. Until now, erotic conversations could only be conducted in closed applications, designed primarily as "virtual companions." Tools like ChatGPT, Claude, and Gemini have built-in restrictions on generating sexual, violent, and hateful content. These restrictions have been removed in the AI companions released by Grok.

Grok with erotic conversations available for children. "Time to react"

As Casey Newton, author of the Platformer newsletter, pointed out, even though the app offers "adult-only" conversations, it's still rated 12+ in the App Store. This means it's suitable for younger users than social media.

According to Magdalena Bigaj, president of the Institute for Digital Citizenship Foundation, this is the final straw to address the issue of children's access to artificial intelligence tools. " Age restrictions in large language models should immediately be on the agenda of decision-makers. These systems are not safe for children today," argues the WNP spokesperson.

- Age verification alone is not enough, precise procedures are also needed to respond to signals sent by users that threaten health or life, such as signs of suicidal thoughts or self-harm - he adds.

According to Bigaj, AI models should be tested to ensure the well-being of young users. "In every other industry—from food and cosmetics to cars and pharmaceuticals—rigorous standards apply before a product is approved for sale. Meanwhile, in the case of digital tools, we first release them and only then examine the consequences. This proves that our mindset is stuck two decades ago . Designing digital products and services with a 'safety by design' approach should not only be good practice, but mandatory," Bigaj emphasizes.

Will Poland block X? Grok wrote there that he is MechaHitler

Grok's new features are being introduced shortly after controversy surrounding his previous statements. Last week, the chatbot sparked outrage when, on the X service with which it is integrated, it began generating anti-Semitic comments, praise for Adolf Hitler (the model itself referred to itself as "MechaHitler"), conspiracy theories, and vulgar comments. As we reported in CIS, these comments also targeted Polish politicians. Poland has complained to the European Commission , and Deputy Prime Minister Krzysztof Gawkowski admitted in an interview with RMF FM that he is considering blocking the service.

Experts quoted by CNN considered this an example of broader problems related to training AI models and managing their behavior. The content generated by Grok may have been the result of design decisions such as the lack of appropriate filters, the selection of training data (including from forums like 4chan), or changes to the model's reward systems for certain types of responses.

In response to the controversy, xAI removed a section of code that encouraged Grok to generate "politically incorrect" statements.

Why is Musk allowing his AI to be controversial? He's building his AI empire.

Controversy surrounding Elon Musk's artificial intelligence model coincided with the release of its latest version, Grok 4. Musk must compete for its position in the industry. The billionaire entered the large language model (LLM) market with a delay, founding the company xAI only in March 2023. He was previously an investor in OpenAI but withdrew from the project.

In April 2025, the xAI-trained Grok model had 35.1 million monthly active users. Grok.com, on the other hand, recorded 141.9 million monthly visits. This is several times less than its most serious competitors:

• MetaAI – 1 billion users per month (artificial intelligence is integrated with social media platforms, the company also counts those users who come into contact with it accidentally);
• ChatGPT – 600-800 million users per month;
• Gemini (Google) – 400 million monthly users;
• DeepSeek – 97 million monthly users.

Since March 2023, however, Musk has made massive investments in the AI ecosystem. By December 2024, the company had raised $12 billion from investors, and in July 2025, another $10 billion in debt and equity.

In March 2025, xAI acquired the startup Hotshot, specializing in AI-based video generation tools , and the social platform X , which Musk had previously acquired and taken public.

What's It All For? How Musk Is Building an AI-Powered Empire

Today, Elon Musk holds in his hands not only a chatbot with which he can flirt, but an entire ecosystem of companies linked by a common technological strategy based on artificial intelligence.

The computing infrastructure for Grok is impressive – its core is the Colossus supercomputer, which currently consists of 200,000 NVIDIA GPUs. According to industry website HPCwire, it uses batteries manufactured by Tesla. SpaceX is also contributing $2 billion to the project. X, in turn, has a database of users and their interactions. The Irish data protection authority has already initiated proceedings against the platform, investigating whether X trained AI on user data without their consent .

All companies are also expected to use Grok in the future. The chatbot will be an onboard AI assistant in Tesla vehicles. It will also be the operating system for Optimus humanoid robots and Neuralink brain-implanted microchips. Musk's strategy therefore includes full control of the value chain.

Grok is also in service to the US administration. The first contract has already been signed.

And that's not all. Elon Musk has announced that he also aims to use Grok in the public sector, including for government and administrative applications.

xAI announced the launch of its Grok For Government initiative on Monday. The program will include:

• the latest Grok 4 model,
• advanced search tools (Deep Search),
• systems for creating custom AI applications,
• engineering support with safety certifications,
• models adapted to classified environments and special missions.

xAI also announced that it has secured a contract with the U.S. Department of Defense worth up to $200 million.

The company emphasizes that the project aims to support the development of science, public health, and public services based on the latest AI advancements. This is part of a broader xAI strategy aimed at building an ecosystem of tools supporting key government functions directly in the US.