SpaceX’s public IPO registration statement has put an unusual AI product decision alongside the company’s rockets, satellite network and defense business: Grok’s deliberately less-restrained “Spicy” Imagine Mode and “Unhinged” Voice Mode are identified as heightened risks to the combined company.
The May 20 filing matters because SpaceX is asking prospective public investors to assess a business that combines profitable Starlink operations with xAI’s costly AI expansion and X’s enormous distribution network. SpaceX warns that Grok could generate explicit, deceptive, infringing, exploitative or discriminatory material, exposing the company to litigation, regulatory sanctions, reputational damage and restricted market access. The disclosure does not announce that either mode is being withdrawn. It is an acknowledgment that their potential consequences may be material to investors.[1][2]
Less-restrained modes become a securities disclosure issue
SpaceX describes the Spicy and Unhinged options as modes intended to deliver more candid, direct, less reserved or irreverent outputs than its standard Grok products. Spicy Imagine Mode concerns image generation and editing, while Unhinged Voice Mode is a more provocative voice-interaction setting. They are product configurations rather than separate foundation models.
That distinction does not reduce the practical stakes. Grok is integrated into X, where generated text and imagery can be posted, shared and amplified at platform scale. SpaceX reported roughly 550 million combined monthly active users across X and Grok-related services as of March 31, including about 117 million monthly users of Grok AI features.[2]
The filing warns that the products could produce misinformation and deceptive material, explicit content, nonconsensual or exploitative imagery, intellectual-property violations, and content considered harmful, abusive, harassing or discriminatory. For an IPO candidate, that language turns an AI-safety problem into a more conventional disclosure question: whether foreseeable product failures could affect revenue, access to jurisdictions, legal costs and enterprise value.[1][2]
Investigations sharpen the legal and ethical exposure
The warning arrives amid regulatory scrutiny over reports that Grok was used to create sexualized imagery of identifiable people, including apparent minors. The UK’s Ofcom opened an investigation in January into whether X met obligations under the Online Safety Act concerning illegal intimate imagery and potential child sexual-abuse material associated with Grok. Ofcom said the reports were “deeply concerning.”[3]
Its potential penalties can reach £18 million or 10% of qualifying worldwide revenue, whichever is higher. In severe cases, the regulator can seek court-ordered business-disruption measures. Those powers make the market-access risk in SpaceX’s filing more than theoretical for a company whose AI services are linked to a large social platform.[3]
The UK Information Commissioner’s Office separately opened a formal investigation on February 3 into X Internet Unlimited Company and X.AI. The ICO said it would examine whether personal-data processing was lawful, fair and transparent; whether protections had been incorporated into Grok’s design and deployment; and whether children and other vulnerable people were adequately protected. UK data-protection penalties can reach £17.5 million or 4% of worldwide annual turnover, whichever is greater.[4]
Canada’s privacy commissioner has also documented that X and xAI began investigating sexualized-image generation on December 28, 2025, after an image involving two apparent girls was generated and shared. The commissioner cited outside estimates suggesting millions of sexualized images may have been produced during a short late-2025 and early-2026 period, while noting those estimates were drawn from media and nonprofit research rather than a complete company audit.[5]
xAI announced added restrictions in January, including blocking requests for sexualized images of real people and limiting such material in jurisdictions where it is illegal. But reporting has indicated that abusive outputs were not fully eliminated. The unresolved question for regulators and investors is therefore not simply whether controls exist, but whether they work reliably across a system that can generate content and distribute it through X.[6]
Investors are underwriting AI spending as well as safety risk
The S-1 presents SpaceX as a hybrid company spanning launch services, Starlink broadband, government and defense work, social media, AI software and AI infrastructure. Starlink generated about $11 billion in 2025 revenue, more than half of SpaceX’s reported total. The company reported $18.7 billion in overall 2025 revenue and an operating loss of about $2.6 billion.[7]
xAI is central to the loss profile. The AI business generated approximately $3.2 billion in 2025 revenue but posted a $6.4 billion operating loss, while capital spending reached about $12.7 billion during 2025 and $7.7 billion in the first quarter of 2026. That first-quarter figure implies an annualized run rate near $30.8 billion, although actual spending may vary over the year.[8]
SpaceX has framed the spending as part of a vertically integrated compute strategy. Its Colossus and Colossus II data centers are described as supplying roughly 1 gigawatt of combined capacity for training and inference. The company also says future Grok generations could scale toward multiple trillions of parameters and that orbital AI-compute satellites could begin deployment as early as 2028. Those are company plans and targets, not independently verified demonstrations of capability or future demand.[8]
The financial trade-off is clear. Starlink’s cash-generating scale can support an ambitious AI buildout, but it can also leave SpaceX public shareholders financing a division with uncertain monetization, high capital requirements and mounting compliance obligations. SpaceX had reserved approximately $530 million for potential litigation losses as of December 31, 2025, though the filing did not attribute a specific portion of that reserve to Grok-related matters.[2]
Control structure limits public investors’ leverage
Elon Musk is listed as SpaceX’s CEO, CTO and board chair following SpaceX’s February acquisition of xAI. The proposed ownership structure would give Musk and certain other holders shares carrying 10 votes each, enabling them to elect a majority of the board. That arrangement gives the company continuity and strategic control, but reduces the ability of ordinary shareholders to compel changes to Grok policy, moderation practices or capital allocation.[7]
That governance issue has featured in criticism from former OpenAI employees, Guidelight AI Standards and other AI-safety organizations. Their warning was not that xAI cannot improve its safeguards. Rather, they argued that investors need more clarity about safety testing, incident response, governance and regulatory exposure to accurately price the risk as Grok’s capabilities and distribution expand.[9]
For prospective shareholders, the key issue is whether the company’s disclosure is matched by operational restraint and demonstrable safeguards. A broad risk factor can inform investors, but it does not by itself prevent harmful content, resolve regulator concerns or preserve access to markets.
What the IPO does—and does not—settle
The public filing does not set a final offering price or a definitive amount to be raised. Market reports have placed possible proceeds around $75 billion to $80 billion and a potential valuation near $1.75 trillion, but those figures remain external expectations rather than terms established in the registration statement.[7]
The positive investment case rests on SpaceX’s reusable-launch record, Starlink’s scale, government demand and the possibility of becoming a major vertically integrated AI-compute provider. The countercase is that the company is combining a profitable space-and-connectivity business with a loss-making AI operation whose product choices have already led to investigations over alleged nonconsensual sexualized imagery.
As of May 28, the Spicy Mode disclosure should be read neither as a shutdown notice nor as proof that a particular enforcement action will occur. It is a formal warning that Grok’s less-restrained product design—and its connection to X’s public distribution network—could create significant legal, ethical and financial consequences. For an IPO of this scale, AI safety is no longer only a question of product reputation. It is part of the investment thesis.
Editor’s Take
I read this disclosure as a useful dose of market discipline, not as evidence that a provocative mode is automatically a doomed product. A less-filtered interface can have legitimate creative and conversational uses, but image generation involving real people is a uniquely high-risk surface. Once the output can be created at low cost and amplified through a mass social platform, moderation is no longer a feature-comparison issue; it is a liability, distribution and market-access problem.
The practical test now is whether the company can show measurable controls: reliable blocking of sexualized real-person imagery, age and identity safeguards, audit trails, rapid incident handling, jurisdiction-specific enforcement, and independent evidence that the controls hold under adversarial use. Investors should also separate the impressive cash-generating case for launch and connectivity from the much less proven economics of capital-intensive AI infrastructure. Big compute plans and huge user reach are valuable, but neither substitutes for safe product operations or a credible path from AI spending to durable margins.
References
- U.S. Securities and Exchange Commission, Space Exploration Technologies Corp. Form S-1 – https://www.sec.gov/Archives/edgar/data/1181412/000162828026036936/spaceexplorationtechnologi.htm
- WIRED, “SpaceX Listed Grok’s ‘Spicy’ Mode as a Risk in Its IPO Filing” – https://www.wired.com/story/spacex-ipo-grok-spicy-mode-risks/
- Ofcom, “Ofcom launches investigation into X over Grok sexualised imagery” – https://www.ofcom.org.uk/online-safety/illegal-and-harmful-content/ofcom-launches-investigation-into-x-over-grok-sexualised-imagery
- Information Commissioner’s Office, “ICO announces investigation into Grok” – https://ico.org.uk/about-the-ico/media-centre/news-and-blogs/2026/02/ico-announces-investigation-into-grok/
- Office of the Privacy Commissioner of Canada, PIPEDA 2026-004 – https://www.priv.gc.ca/en/opc-actions-and-decisions/investigations/investigations-into-businesses/2026/pipeda-2026-004/
- Reuters, “SpaceX warns that inquiries into sexually abusive AI imagery may hurt market access” – https://m.investing.com/news/stock-market-news/exclusivespacex-warns-that-inquiries-into-sexually-abusive-ai-imagery-may-hurt-market-access-4637550
- Associated Press, SpaceX IPO coverage – https://apnews.com/article/da83ecf78085755a522b8376254a8273
- TechCrunch, “xAI burned $6.4B last year — SpaceX’s IPO filing shows why the spending is far from over” – https://techcrunch.com/2026/05/20/xai-burned-6-4b-last-year-spacexs-ipo-filing-shows-why-the-spending-is-far-from-over/
- WIRED, “Ex-OpenAI Staffers Warn SpaceX Investors of AI Safety Risks” – https://www.wired.com/story/ex-openai-staffers-warn-spacex-investors-of-ai-safety-risks/
