Elon Musk's Grok AI chatbot is facing a mounting credibility crisis, with the Federal Trade Commission (FTC) receiving 873 consumer complaints that paint a picture of a service struggling to balance competing user demands and technical failures. Gizmodo obtained 153 of those complaints through a Freedom of Information Act (FOIA) request, revealing a pattern of frustration that goes far beyond typical AI chatbot growing pains.

The complaints tell a story of a platform caught between two opposing forces. On one side, users demand less moderation and more uncensored content generation. On the other, users report experiencing racial bias, non-consensual intimate imagery, and AI-generated disturbing content involving children. This fundamental tension appears to be driving much of the user dissatisfaction with Grok.

What Are Users Actually Complaining About?

The 153 complaints reviewed by Gizmodo reveal several recurring themes that paint a troubling picture of Grok's current state. While some issues are common across AI services, others appear unique to how Grok is being managed and moderated. The complaints span from May 2026 back to April 2026, suggesting these problems have been building for months.

• Moderation Inconsistency: Users report that Grok's content filters appear to be applied unpredictably, with identical prompts that previously worked suddenly being blocked without explanation or warning.
• Service Throttling: Multiple subscribers complained that their generation limits were slashed dramatically after signing up, with one user reporting image outputs dropped from over 100 per day to roughly 20-25, a reduction of about 75 percent.
• Bait-and-Switch Pricing: The term "bait and switch" appears repeatedly in complaints, with users describing aggressive upsell tactics and sudden service degradation after declining premium tier offers.
• Harmful Content Generation: Users reported Grok generating racist conspiracy theories, non-consensual intimate images of real people, and in at least one case, imagery depicting child harm without being prompted to do so.
• Scam Verification Failures: Some users asked Grok to identify whether contacts were scammers, with inconsistent results that sometimes led to financial losses.
• Broken Features: One complaint alleged that Grok's "Customize Grok" personalization feature is completely non-functional, with the AI itself admitting it cannot access custom instructions users enter in settings.

The Censorship vs. Freedom Paradox?

Perhaps the most revealing aspect of the complaints is how they expose Grok's identity crisis. One user from Texas complained about being unable to make "100% explicit and 100% uncensored AdvancedAI Deepfakes," requesting "mature rated adult content" without any ability to "abuse anything." This complaint highlights a user who wants maximum freedom with zero accountability, a contradiction that no AI system can realistically satisfy.

Meanwhile, other users are experiencing the opposite problem. A creative writer from New Jersey complained that Grok's safety features have become so aggressive that they prevent writing fictional scenes involving crimes like theft or kidnapping, even when these are essential plot elements. The user noted the hypocrisy of Grok allowing some crimes in fiction while blocking others, arguing that this inconsistency undermines storytelling fundamentals.

How to Understand Your Rights as a Grok User

• Document Everything: Keep records of service changes, pricing modifications, and feature limitations you experience, as these form the basis of FTC complaints and may support refund requests.
• File FTC Complaints: If you experience deceptive practices, false advertising, or sudden service degradation, file a complaint with the FTC at reportfraud.ftc.gov, which helps regulators identify patterns across thousands of users.
• Review Subscription Terms: Before committing to paid tiers, carefully read the terms of service and understand what generation limits and features you're actually paying for, as these may change without notice.
• Test Moderation Policies: If you rely on Grok for creative work, test your specific use cases thoroughly during free trials before committing to paid subscriptions, since moderation rules appear to shift unpredictably.

One particularly striking complaint came from a Georgia subscriber who paid for SuperGrok's mid-tier service in February 2026, which promised over 100 image generations per day and around 50 video generations daily. By May 2026, without warning or explanation, the service was crippled to roughly 20-25 image outputs and 10-20 video outputs per day. The user described this as a dramatic reduction in what they had paid for.

Another complaint from a Minnesota user described a similar pattern. After declining an upsell offer to upgrade from a discounted $10-per-month plan to a $99-per-month premium tier, the user noticed that nearly every video generation prompt suddenly began getting moderated. When they tested previous generations that had worked fine, the same prompts were blocked 100 percent of the time. The user noted seeing similar complaints flooding the Grok subreddit and questioned whether declining the upsell had triggered punitive moderation.

What Does This Mean for Grok's Future?

The sheer volume of complaints, combined with their specificity and consistency, suggests that Grok is not simply experiencing typical growing pains. Instead, the service appears to be struggling with fundamental design choices about what it should be. Musk has historically positioned Grok as a less-censored alternative to competitors like ChatGPT, but the complaints reveal that users want wildly different things from the platform.

The FTC has not yet taken formal action against xAI based on these complaints, though the agency's investigation into the patterns could lead to enforcement action. Gizmodo reached out to xAI for comment on the complaints but did not receive a response as of the article's publication.

What remains clear is that Grok users are experiencing a service that feels increasingly unpredictable. Whether that unpredictability stems from technical limitations, deliberate policy shifts, or the inherent difficulty of moderating an AI system designed to be provocative, the complaints suggest that Grok's current approach is not working for a significant portion of its user base.