Elon Musk is making expansive claims for the latest version of his AI model.
This week xAI, the artificial intelligence startup founded by Elon Musk, revealed the latest iteration of its language model, Grok, branding it as the “smartest” and “maximally truth-seeking” AI available. The announcement was made during a livestreamed demonstration on X, Musk’s social media platform.
Musk claimed that Grok-3, which has been trained with ten times more computing power than its predecessor, surpasses competing models, including OpenAI’s GPT-4o and DeepSeek’s V3, in key performance metrics such as mathematics, expert reasoning and coding abilities.
A significant feature of Grok-3 is its integration with X, formerly Twitter, enabling it to access and generate responses based on content posted by X users. Musk argues this makes X a credible source for news, aligning with his broader vision to leverage AI for real-time news synthesis from social media.
Musk’s critical stance on mainstream journalism has been evident, amplified by recent posts suggesting journalists should face imprisonment and asserting that X users are the new media.
Since last year, Grok has been generating summaries of trending topics based on popular posts on X, effectively capturing public sentiment. However, this approach has also led to the dissemination of misinformation, with examples including false narratives about international events. In response to concerns about inaccuracies, particularly regarding electoral information, Musk’s company implemented updates to include links to official resources like Vote.gov.
Differentiating itself from competitors such as OpenAI, which have established partnerships with news organisations for content licensing, xAI has not pursued formal agreements with publishers. Musk’s recent rhetoric, reflecting some antagonism towards established media, was highlighted when he shared a Grok-3 response labelling the tech news site The Information as “garbage,” while simultaneously praising X as an unmediated source of news straight from users. Follow-up tests by NBC News on Grok-3 yielded a more balanced description of The Information.
A recent analysis by Marina Adami at the Reuters Institute for the Study of Journalism suggests that Grok-2, the model preceding Grok-3, primarily sourced its information from reputable official and nonpartisan news sources when responding to queries about German elections. While the overall findings indicated a balanced output, she noted a trend towards right-leaning content influenced by Musk.
To evaluate Grok-3’s sourcing capabilities, a controlled experiment was conducted, prompting Grok-2 and Grok-3 with quotes from 200 articles across 20 news organisations. The results showed Grok-3 struggled with accurate publisher attribution, identifying the correct source only 21% of the time and delivering accurate responses only five times out of 200 instances. Furthermore, it frequently provided citations that, while appearing valid, led to error messages and fabricated links, raising questions about the reliability and integrity of the information generated.
Both Grok iterations demonstrated a tendency to “hallucinate,” meaning they produce plausible-sounding yet false information, a trait that can mislead users, particularly when querying news-related topics. An investigation conducted by BBC News mirrored these findings, revealing substantial inaccuracies in responses from several chatbots when tasked with news questions, with many containing incorrect facts or quotations. Industry sources have emphasised the necessity for news organisations to control the use and representation of their content by AI systems, deeming it crucial for maintaining trustworthiness.
Source: Noah Wire Services
- https://opencv.org/blog/grok-3 – This URL supports the claim about Grok-3’s advanced AI capabilities, including its improved performance and accuracy compared to its predecessors and other AI models like ChatGPT and DeepSeek.
- https://x.ai/blog/grok-3 – This URL provides details on Grok 3’s performance across various benchmarks and its integration with the X platform, aligning with Elon Musk’s vision for real-time news synthesis.
- https://www.noahwire.com – This URL is the source of the article discussing Grok-3’s features and implications for journalism, including its integration with X and the challenges of AI-driven news synthesis.
- https://www.reuters.com – Reuters has reported on Grok-3’s capabilities and performance, which can be related to its role in news synthesis and the broader context of AI in journalism.
- https://www.bbc.com/news – BBC News has investigated the accuracy of chatbots, including Grok iterations, highlighting issues with misinformation and the tendency to ‘hallucinate’ false information.
Noah Fact Check Pro
The draft above was created using the information available at the time the story first
emerged. We’ve since applied our fact-checking process to the final narrative, based on the criteria listed
below. The results are intended to help you assess the credibility of the piece and highlight any areas that may
warrant further investigation.
Freshness check
Score:
8
Notes:
The narrative references recent developments and ongoing events, such as Elon Musk’s activities and the evolution of AI in journalism. However, specific dates or timelines for some events are not provided, which could slightly reduce its freshness.
Quotes check
Score:
6
Notes:
There are no direct quotes from specific individuals that can be verified online. The narrative does mention Musk’s general views and statements but lacks specific, attributable quotes.
Source reliability
Score:
9
Notes:
The narrative originates from a reputable media outlet, Columbia Journalism Review (CJR), which is known for its in-depth analysis of media and journalism. References to other credible sources like the Reuters Institute and BBC News further enhance reliability.
Plausability check
Score:
8
Notes:
The claims about Grok-3 and its capabilities are plausible given the current advancements in AI technology. However, the narrative highlights challenges with accuracy and sourcing, which are consistent with known issues in AI-generated content.
Overall assessment
Verdict (FAIL, OPEN, PASS): PASS
Confidence (LOW, MEDIUM, HIGH): HIGH
Summary:
The narrative is well-supported by references to recent events and credible sources, enhancing its reliability. While specific quotes are not verified, the overall plausibility of the claims aligns with current trends in AI and journalism. The freshness of the content is also high, given its relevance to ongoing developments.
Register for Editor’s picks
Stay ahead of the curve with our Editor's picks newsletter – your weekly insight into the trends, challenges, and innovations driving the future of digital media.
