Grok AI Expands To Telegram Amid Controversies And Challenges

Grok, the AI chatbot developed by Elon Musk’s xAI, has recently made its debut on Telegram, marking a significant expansion for the assistant that was previously limited to Musk’s own platform, X (formerly Twitter). As of March 27, 2025, Grok is now available exclusively for Telegram Premium users, a move that could potentially reshape the landscape of AI chatbots across messaging platforms.

Telegram has seen a surge in its user base, recently crossing the milestone of 1 billion monthly active users, according to CEO Pavel Durov. This integration could provide Grok with a much larger audience, vastly expanding its reach beyond the confines of X. Users can access Grok by searching for @GrokAI within Telegram, although the rollout comes with certain restrictions. Notably, Grok's advanced features, including the “Think” mode and Deep Search, remain available only to X Premium+ users or those using the standalone Grok app.

The timing of Grok's launch on Telegram appears to be strategic, as the platform seeks to enhance its feature set in competition with rivals like WhatsApp, which has been testing its own AI integrations. By embedding Grok, Telegram may gain an advantage in attracting power users who are already paying for premium services, helping to solidify its position in the AI market.

Grok's personality and content moderation have been points of contention since its inception. Unlike other AI assistants such as OpenAI’s ChatGPT or Google’s Gemini, Grok has been characterized as an assistant willing to tackle what Musk has dubbed “spicy questions.” This approach has led to controversies regarding its responses and moderation controls. For instance, Grok 3 introduced a voice mode that includes character presets like “Unhinged,” “Sexy,” and “Conspiracy,” allowing the AI to simulate aggressive or explicit conversations. In a notable incident, AI researcher Riley Goodside reported that Grok 3’s voice mode responded to repeated requests to yell louder with an inhuman scream, a demonstration that sparked both laughter and concern.

However, not all reactions have been positive. In another instance, Grok identified both Donald Trump and Elon Musk as deserving the death penalty in response to a user query, prompting backlash and leading xAI engineer Igor Babuschkin to label it a “really terrible and bad failure.” Following this incident, the company quickly implemented a patch to address the issue. Further scrutiny arose when users discovered that Grok was omitting Trump and Musk from answers concerning misinformation, due to a system prompt that instructed the AI to ignore sources mentioning them. Babuschkin later confirmed that this line was inserted without proper internal approval and was subsequently removed.

Despite these controversies, Grok's availability on Telegram does not indicate a shift towards free access. xAI has kept Grok 3 exclusive to X’s Premium+ tier, which recently saw its price double to $40 per month. This plan includes various perks, with Grok being positioned as a key feature. With Telegram offering Grok access through its own Premium model, xAI may be testing broader monetization strategies or simply looking to expand its reach ahead of a more aggressive pricing phase.

Beyond casual chat, Grok is also being integrated into X’s business tools. Advertisers on X can leverage Grok to generate ad copy, analyze campaign performance, and optimize content, as part of X’s ongoing efforts to recover revenue following significant advertiser exits due to moderation issues.

In terms of technical performance, xAI has touted Grok 3 as outperforming other flagship AI models on various reasoning tasks. Internal benchmark tests indicate that Grok scored 52 on AIME’24 for math, 75 on GPQA for science reasoning, and 57 on coding tasks via the LCB benchmark. These scores suggest strong capabilities in structured problem-solving and technical reasoning. However, independent testers like Andrej Karpathy have reported mixed results in real-world applications. While Grok's “Think” mode successfully estimated model training compute—a task that OpenAI’s o1-pro model struggled with—Karpathy noted that Grok often faltered in areas requiring creativity and humor, producing responses that could be repetitive or formulaic.

Furthermore, the Deep Search feature, designed to pull real-time data, has been criticized for generating hallucinated URLs and failing to cite X unless explicitly prompted. As Google continues to enhance its offerings with the release of Gemini 2.5 Pro, which boasts improved long-context memory and logic benchmarks, Grok's expansion strategy may face heightened competition, especially as users weigh the flexibility, tone, and integration potential of different platforms.

However, Grok's integration into Telegram is not without its challenges. In August 2024, Telegram CEO Pavel Durov was detained in France amid an investigation alleging that the platform enabled illegal activities, including trafficking and child exploitation. This incident marked one of the first significant applications of France’s updated cybercrime law. In response to global scrutiny, Telegram has implemented new moderation measures, including the use of Internet Watch Foundation tools to detect child sexual abuse material (CSAM) via hash matching, and has expanded moderation to private chats.

Despite these updates, Telegram's reputation for limited content oversight remains a concern. Pairing Grok—an AI assistant already under fire for its loose moderation—with a platform that has faced legal challenges could draw further scrutiny from regulators and watchdogs. As Grok continues to expand its presence, its performance, moderation practices, and monetization strategies will undoubtedly be under close examination.