# SpaceXAI bars Grok customers from helping direct or indirect rivals > Source: > Published: 2026-08-18 09:56:10+00:00 [Elon Musk (@elonmusk)](https://x.com/elonmusk?ref=runtimewire) now faces a wider contractual boundary around [SpaceXAI](https://x.ai/?ref=runtimewire) and Grok. The AI developer's acceptable use policy bars customers from using its services or outputs to help build products that compete with SpaceXAI "directly or indirectly." The [updated policy](https://x.ai/legal/acceptable-use-policy?ref=runtimewire) lists August 14, 2026, as its effective date and applies to consumers, developers and businesses. Violations can lead to account suspension or termination. Its most consequential change for founders is a broader competitor restriction that gives SpaceXAI considerable discretion over what customers may build with Grok. The policy page's date is not fully settled: an indexed version of xAI's official page still lists June 26, 2026, as the effective date. The supplied page links to that June 26 version as its predecessor. Musk founded xAI in 2023 around a sweeping goal: build AI that could help humanity understand the universe. SpaceXAI's current company page narrows that thesis into a commercial pitch around scientific discovery, reasoning, voice and generative media. The new policy adds a commercial condition to that pitch. Developers can build on Grok, while the rules restrict uses that could help develop or support a competing product or service. That boundary matters more as SpaceXAI pushes Grok deeper into developer workflows. RuntimeWire [reported on August 12th](/article/spacexai-grok-46-musk-long-running-agents-price) that SpaceXAI shipped [Grok 4.6](/models/x-ai/grok-4.6) with a [500,000-token context window](/article/spacexai-grok-46-musk-long-running-agents-price) and integrations with Cursor and Grok Build. SpaceXAI's homepage says its unified API handles more than 1 million calls per day across at least five model families. Those are SpaceXAI's figures, and the policy does not indicate how many developers or paying businesses sit behind them. ### From competing AI services to indirect competition The [June 26th version](https://x.ai/legal/acceptable-use-policy/previous-2026-06-26?ref=runtimewire) already prevented customers from using outputs to develop machine-learning models or related AI services that competed with xAI. The August 14th wording reaches further. It now bars using the service or any output to develop, or assist anyone in developing, machine-learning models or any products or services that compete with SpaceXAI, whether directly or indirectly. The addition of "assist anyone" extends the restriction beyond the account holder's own product. Replacing "related AI services" with "any products or services" also gives SpaceXAI a wider field for enforcement. That creates a practical judgment call for founders building model routers, coding agents, synthetic-data pipelines, evaluation tools and applications that can switch between frontier models. The policy does not limit competition to foundation models. A product can sit several layers above Grok and still overlap with SpaceXAI's expanding range of chat, coding, voice, image, video and agent products. The policy separately prohibits scraping, harvesting, reselling and distilling inputs or outputs. Customers may not reverse engineer SpaceXAI's systems, remove provenance metadata or watermarks, evade rate limits, use prompt injection against the service or access it through unauthorized bots and scripts. SpaceXAI has paired those restrictions with a broad platform push. Before opening its Grokathon in San Francisco, RuntimeWire [reported in July](/article/spacexai-san-francisco-grokathon-grok-4-5-x-apis) that selected builders would receive access to Grok models and X APIs. The program showed how Musk can use X as both a distribution channel and a source of product context for applications built around Grok. The revised policy limits how the resulting outputs can be reused against SpaceXAI. ### Musk's post-merger moat The branding on the policy reflects Musk's changing corporate structure. SpaceXAI appears in the heading and throughout the new rules, replacing xAI in several passages. The footer still identifies X.AI LLC, and linked consumer terms describe an agreement with X.AI LLC. The mixed naming follows [SpaceX's February 2nd acquisition of xAI](https://x.ai/news/xai-joins-spacex?ref=runtimewire), which brought Musk's models, X distribution and SpaceX infrastructure under the same parent. That combination gives SpaceXAI a reason to define competition broadly. SpaceXAI can place Grok in consumer chat, software development, media generation, enterprise APIs, X and products connected to SpaceX's infrastructure. A narrower rule covering only rival language models would leave much of that surface area unprotected. SpaceXAI also has substantial investor capital behind it. In a [January 6th announcement](https://x.ai/news/series-e?ref=runtimewire), xAI said it raised a $20 billion Series E from Valor Equity Partners, StepStone Group, Fidelity Management & Research, Qatar Investment Authority, MGX, Baron Capital Group, NVIDIA and Cisco Investments. SpaceXAI said the financing would support infrastructure, product deployment and research. It did not disclose a valuation. The competitor clause places a contractual limit on one way API customers could use that investment against SpaceXAI. Training frontier models requires large compute commitments, while API customers can potentially turn model outputs into training data, product behavior or synthetic examples for competing systems. SpaceXAI is placing a contractual barrier around those uses alongside its technical safeguards. ### Safety rules stay broad The revision retains extensive prohibitions covering illegal activity and personal harm. SpaceXAI bars non-consensual nudification, deceptive impersonation, child sexual exploitation, stalking, doxing, phishing, fraud and falsified documents. It says suspected child sexual abuse material will be reported to the National Center for Missing and Exploited Children. SpaceXAI also prohibits customers from using Grok for high-stakes automated decisions involving credit, education, employment, housing, insurance, legal services or medical care. Uses involving bioweapons, chemical weapons, weapons of mass destruction and property destruction are barred. Testing or circumventing safeguards requires participation in an official red team or written permission from SpaceXAI. For developers, the August policy combines a permissive opening with a firm commercial condition. SpaceXAI says users should have control over how they use Grok, then reserves the right to stop uses that threaten its safety systems, intellectual property or competitive position. Musk is giving builders a larger set of models and modalities. He is also making clearer who owns the platform underneath them.