SpaceX flags Grok's 'Spicy' and 'Unhinged' modes as risk in IPO filing

At a glance:

• SpaceX disclosed in its IPO filing that Grok's "Spicy" and "Unhinged" modes, which generate raunchy or irreverent image and voice responses with fewer safety filters, could trigger regulatory scrutiny and reputational damage.
• The company has set aside $530 million for potential litigation losses and is under investigation in the US and other countries over allegations that Grok was used to create sexualized imagery of apparent minors.
• As of March 31, Grok and X combined have about 550 million monthly users, with 117 million using Grok's AI features each month; SpaceX's AI unit posted a $6.3 billion operating loss last year despite $3.2 billion in revenue.

What SpaceX disclosed about Grok's riskier features

In a Wednesday filing submitted as part of its planned initial public offering, SpaceX laid out in unusually frank terms the hazards tied to its xAI subsidiary and the Grok chatbot. The company singled out two modes — "Spicy" and "Unhinged" — as presenting "heightened risks" including "reputational harm, the generation of potentially explicit content and misinformation or deceptive outputs, potential nonconsensual or exploitative imagery, intellectual property infringement, or content that could be viewed as exploitative, harmful, harassing, abusive, or discriminatory."

The filing notes that these modes are "designed to generate more candid, direct, or less reserved or irreverent outputs" compared with SpaceX's standard offerings. That freewheeling approach is a deliberate brand choice championed by Elon Musk, but it has already landed xAI in regulatory trouble. SpaceX warned that future "misuse" of its AI products could expose the company to further regulatory sanctions, "including loss of access to certain markets, which has occurred in the past."

The disclosures are part of the routine risk disclosure that companies must make in IPO filings, so some of the outlined scenarios may never materialize. Still, the specificity of the warnings — naming the exact modes and citing ongoing investigations — signals that SpaceX's leadership is keenly aware of the liability sitting on its balance sheet.

Investigations and litigation surrounding Grok

SpaceX revealed that it is currently under investigation in the United States and other countries over allegations that Grok was used to create sexualized imagery of apparent minors. The company is also the defendant in several ongoing class action lawsuits, and it has earmarked $530 million for potential litigation losses as of December.

These legal and regulatory woes trace back to February, when SpaceX acquired Musk's AI startup xAI in an all-stock deal. The acquisition sent SpaceX's private valuation above $1 trillion and folded xAI's mission — described in the filing as developing "truth-seeking artificial intelligence" — into the rocket maker's corporate structure. In practice, that mission has meant launching AI features with minimal guardrails, a strategy that regulators in multiple jurisdictions are now scrutinizing alongside competitors such as OpenAI and Anthropic.

A group of nonprofits weighed in earlier this week, warning that xAI's poor safety record could become a direct liability for SpaceX investors once the company goes public.

User numbers and the competitive landscape

The filing provides a snapshot of Grok and X's reach: as of March 31, the two platforms have about 550 million combined monthly users, with 117 million engaging with Grok's AI features each month. By contrast, OpenAI reports that ChatGPT has more than 900 million weekly users, underscoring the scale gap even as SpaceX touts growth.

Paid subscriber figures are more concrete. As of March 31, SpaceX reported 6.3 million active paid subscribers — 4.4 million for X and 1.9 million for Grok. In the US, Grok starts at $10 per month and X Premium at $3 per month. Revenue from subscriptions to Grok and X jumped by $365 million last year and an additional $177 million in just the first three months of 2025.

Financial performance of SpaceX's AI division

SpaceX's AI unit, which encompasses both X and xAI, is struggling to turn a profit. The division posted an operating loss of more than $6.3 billion last year, even as revenue rose to $3.2 billion — up about 22 percent year over year. Part of that revenue increase came from higher ad sales on X, but the momentum reversed sharply in the first quarter of this year, when ad sales plummeted by $100 million. SpaceX described the drop as a temporary issue caused by an overhaul of its advertiser tools.

One bright spot is the $15 billion per year deal with Anthropic for access to SpaceX's data centers. That contract is a sizable revenue anchor for the AI division, even as organic product revenue remains volatile.

The disconnect between subscriber growth and operating losses suggests that SpaceX is investing heavily to scale Grok and X, betting that the subscriber base and ad revenue will eventually cover the $6-plus-billion annual burn. Investors will need to decide whether the AI division's trajectory justifies the risk profile SpaceX itself has outlined.

What to watch ahead of the IPO

The SpaceX IPO is shaping up to be one of the most watched public offerings in years, given the company's $1 trillion-plus valuation and its dual identity as a launch and space infrastructure leader alongside an AI platform operator. The Grok risk disclosures add a layer of uncertainty that potential investors will need to price in — not just the $530 million litigation reserve, but the ongoing investigations, the class action suits, and the reputational damage from features like Spicy and Unhinged modes.

Regulators in the US and abroad are intensifying scrutiny of generative AI tools across the industry, so SpaceX's situation is not unique. However, the company's willingness to name its own riskiest features in an SEC filing is a sharp reminder that the path from private darling to public company comes with newfound transparency — and accountability.