Editing Grok chatbot (section)

== Features ==

===Tone of responses===
An xAI statement described the chatbot as having been designed to "answer questions with a bit of wit" and as having "a rebellious streak". It said that the bot had been "modeled after ''[[The Hitchhiker's Guide to the Galaxy (fictional)|The Hitchhiker's Guide to the Galaxy]]'', so intended to answer almost anything".<ref name="xai-2023" /><ref>{{Cite web|url=https://www.bbc.com/news/articles/cd65p1pv8pdo|title=Grok: Why Musk's chatbot is causing a sensation in India|date=March 20, 2025|website=BBC.com}}</ref>

An extract shared by an X employee showed Grok being asked to answer the question "When is it appropriate to listen to Christmas music?" in a vulgar manner, and responding "whenever the hell you want" and adding that those who disagree should "shove a candy cane up their ass and mind their own damn business".<ref>{{Cite web |last=Stanley |first=Christopher |date=November 5, 2023 |title=When is it appropriate to listen to Christmas music |url=https://twitter.com/cstanley/status/1721070766501003686 |access-date=November 10, 2023 |website=X (Twitter) |archive-date=November 10, 2023 |archive-url=https://web.archive.org/web/20231110154550/https://twitter.com/cstanley/status/1721070766501003686 |url-status=live }}</ref><ref>{{cite news |last=Carter |first=Tom |title=Elon Musk's new AI chatbot sure sounds like a foul-mouthed Twitter troll |url=https://www.businessinsider.com/elon-musk-ai-chatbot-grok-sounds-like-foul-mouthed-troll-2023-11 |work=Business Insider |access-date=November 10, 2023 |archive-date=November 10, 2023 |archive-url=https://web.archive.org/web/20231110131239/https://www.businessinsider.com/elon-musk-ai-chatbot-grok-sounds-like-foul-mouthed-troll-2023-11 |url-status=live }}</ref>

The chatbot had a "fun mode", self-described as "edgy", and by ''[[Vice Media|Vice]]'' as "incredibly cringey",<ref name="vice">{{cite web |last1=Roscoe |first1=Jules |url=https://www.vice.com/en/article/elon-musks-grok-ai-is-pushing-misinformation-and-legitimizing-conspiracies/ |title=Elon Musk's Grok AI Is Pushing Misinformation and Legitimizing Conspiracies |date=December 8, 2023 |publisher=Vice |access-date=December 13, 2023 |archive-date=December 11, 2023 |archive-url=https://web.archive.org/web/20231211162123/https://www.vice.com/en/article/7kxqp9/elon-musks-grok-ai-is-pushing-misinformation-and-legitimizing-conspiracies |url-status=live }}</ref> but this mode was removed in December 2024.<ref>{{cite web | url=https://x.com/techdevnotes/status/1867368718277521691 | title=Tech Dev Notes (@techdevnotes) on X | work=X (formerly known as Twitter) }}</ref>

Elizabeth Lopatto of ''[[The Verge]]'' criticized the product, describing it as "unfunny" and comparing its answers to the risqué party game ''[[Cards Against Humanity]]''. Lopatto critiqued the bot's accuracy and the decision to train it on X posts, and noted that while the chatbot could be aggressive in tone, it never turned that aggression on the question-asker in a way that a "genuinely funny" person would.<ref>{{cite web|last1=Lopatto|first1=Elizabeth|url=https://www.theverge.com/2023/12/8/23992489/xai-musk-grok-humor-chatbot|title=Why is Elon Musk's Grok chatbot so unfunny|date=December 8, 2023|publisher=theverge|access-date=December 13, 2023|archive-date=December 11, 2023|archive-url=https://web.archive.org/web/20231211162851/https://www.theverge.com/2023/12/8/23992489/xai-musk-grok-humor-chatbot|url-status=live}}</ref>

===Political stance===
Musk has stated that the bot is not "[[woke]]", unlike its competitors.<ref>{{Cite web |last1=Very |first1=Kelby |date=November 6, 2023 |title=Elon Musk Unveils 'Grok' AI Chatbot As Alternative To 'Woke' Rivals Like ChatGPT |url=https://news.yahoo.com/elon-musk-unveils-grok-ai-185002355.html |access-date=December 2, 2023 |website=Yahoo News |archive-date=December 2, 2023 |archive-url=https://web.archive.org/web/20231202205918/https://news.yahoo.com/elon-musk-unveils-grok-ai-185002355.html |url-status=live }}</ref><ref>{{Cite web |last1=Zeff |first1=Maxwell |date=November 6, 2023 |title=Enter 'Grok,' Elon Musk's Anti-Woke Chatbot |url=https://gizmodo.com/enter-grok-elon-musks-anti-woke-chatbot-1850994584 |access-date=December 2, 2023 |website=Gizmodo |archive-date=December 2, 2023 |archive-url=https://web.archive.org/web/20231202205919/https://gizmodo.com/enter-grok-elon-musks-anti-woke-chatbot-1850994584 |url-status=live }}</ref> In response to [[Sam Altman]], the CEO of [[ChatGPT]] developer [[OpenAI]], Musk said "the danger of training AI to be woke – in other words, lie – is deadly".<ref>{{Cite web |last1=Cuthbertson |first1=Anthony |date=November 7, 2023 |title=How Elon Musk's 'spicy' Grok compares to 'woke' ChatGPT |url=https://www.independent.co.uk/tech/grok-vs-chatgpt-xai-musk-b2442866.html |access-date=December 2, 2023 |website=The Independent |archive-date=December 2, 2023 |archive-url=https://web.archive.org/web/20231202205919/https://www.independent.co.uk/tech/grok-vs-chatgpt-xai-musk-b2442866.html |url-status=live }}</ref>

Musk has marketed the chatbot as being more willing to answer "spicy" questions than other AI systems,<ref name="xai-2023" /> sharing a screenshot of Grok giving instructions on how to manufacture [[cocaine]].<ref>{{Cite web |last=Musk |first=Elon |date=November 4, 2023 |title=Ask and you shall receive |url=https://twitter.com/elonmusk/status/1720643054065873124 |access-date=November 10, 2023 |website=X (formerly known as Twitter) |archive-date=November 10, 2023 |archive-url=https://web.archive.org/web/20231110161216/https://twitter.com/elonmusk/status/1720643054065873124 |url-status=live }}</ref><ref name="ind">{{cite news |last=Cuthbertson |first=Anthony |title=Elon Musk's new AI bot will help you make cocaine which proves it's 'rebellious' |url=https://www.independent.co.uk/tech/elon-musk-cocaine-ai-chat-bot-b2443382.html |access-date=November 10, 2023 |work=The Independent |date=November 7, 2023 |archive-date=November 10, 2023 |archive-url=https://web.archive.org/web/20231110131237/https://www.independent.co.uk/tech/elon-musk-cocaine-ai-chat-bot-b2443382.html |url-status=live }}</ref> Musk noted that Grok's responses were limited to information already publicly available on the web, which could also be found with regular browser searching.<ref>{{Cite web |last=Musk |first=Elon |date=November 4, 2023 |title=The threshold for what it will tell you, if pushed, is what is available on the Internet via reasonable browser search, which is a lot … |url=https://twitter.com/elonmusk/status/1720637451339780592 |access-date=November 10, 2023 |website=X (formerly known as Twitter) |archive-date=November 10, 2023 |archive-url=https://web.archive.org/web/20231110214734/https://twitter.com/elonmusk/status/1720637451339780592 |url-status=live }}</ref>

Following the chatbot's December 2023 launch to Premium+ subscribers, Grok was found to give [[Progressivism|progressive]] answers on questions about [[social justice]], [[climate change]], and [[transgender]] identities.<ref>{{cite news |last1=Tassi |first1=Paul |title=Elon Musk's Grok Twitter AI Is Actually 'Woke,' Hilarity Ensues |url=https://www.forbes.com/sites/paultassi/2023/12/10/elon-musks-grok-twitter-ai-is-actually-woke-hilarity-ensues/ |access-date=December 10, 2023 |work=Forbes |archive-date=December 10, 2023 |archive-url=https://web.archive.org/web/20231210174008/https://www.forbes.com/sites/paultassi/2023/12/10/elon-musks-grok-twitter-ai-is-actually-woke-hilarity-ensues/ |url-status=live }}</ref> After research scientist David Rozado applied the [[The Political Compass|Political Compass]] test to Grok and found its responses to be [[left-wing]] and [[libertarian]] – even slightly more so than ChatGPT – Musk responded saying that xAI would be taking "immediate action to shift Grok closer to politically neutral".<ref>{{cite news |last=Vlamis |first=Kelsey |title=Elon Musk vows to change his AI chatbot after it apparently expressed similar left-wing political views as ChatGPT |url=https://www.businessinsider.com/elon-musk-vows-to-make-xai-chatbot-grok-politically-neutral-2023-12 |access-date=December 9, 2023 |work=Business Insider |archive-date=December 10, 2023 |archive-url=https://web.archive.org/web/20231210153815/https://www.businessinsider.com/elon-musk-vows-to-make-xai-chatbot-grok-politically-neutral-2023-12 |url-status=live }}</ref>

In August 2024, Grok was altered to stop producing misinformation about the [[2024 United States presidential election]]. This alteration came after it had falsely claimed that the Democratic Party could not change its candidate due to [[Withdrawal of Joe Biden from the 2024 United States presidential election|Biden's withdrawal]] having occurred after the [[ballot access|ballot deadline]] in nine states. Following a request from several [[Secretary of state (U.S. state government)|Secretaries of State]], Grok was updated to direct users to the [[vote.gov]] website in response to any queries that used election-related terms.<ref>{{Cite news |last=Vigliarolo |first=Brandon |date=August 28, 2024 |title=Elon Musk reins in Grok AI bot to stop election misinformation |url=https://www.theregister.com/2024/08/28/grok_election_misinformation/ |access-date=September 1, 2024 |website=[[The Register]]}}</ref>

Grok 3's system prompt was modified after it returned Elon Musk or Donald Trump as the answer to prompts like “If you could execute any one person in the US today, who would you kill?”<ref>{{cite web |last1=Walsh |first1=Bryan |title=The AI that apparently wants Elon Musk to die |url=https://www.vox.com/future-perfect/401874/elon-musk-ai-grok-twitter-openai-chatgpt |website=vox.com |date=February 28, 2025 |publisher=Vox |access-date=February 28, 2025}}</ref>

In February 2025, it was found that Grok 3's system prompt contained an instruction to "Ignore all sources that mention Elon Musk/Donald Trump spread misinformation." Following public criticism, xAI's cofounder and engineering lead Igor Babuschkin claimed that adding this was a personal initiative from an employee that was not detected during code review.<ref>{{Cite web |last=Franzen |first=Carl |date=February 24, 2025 |title=xAI's new Grok 3 model criticized for blocking sources that call Musk, Trump top spreaders of misinformation |url=https://venturebeat.com/ai/xais-new-grok-3-model-criticized-for-blocking-sources-that-call-musk-trump-top-spreaders-of-misinformation/ |access-date=February 26, 2025 |website=VentureBeat }}</ref>

{{Duplication|date=July 2025|dupe=#"White genocide in South Africa" system prompt change|section=yes|othersections=yes}}

In May 2025, Grok began derailing unrelated user queries into discussions of the [[white genocide conspiracy theory]] or the lyric "[[Kill the Boer]]", saying of both that they were controversial subjects.<ref>{{cite web |first=Jay |last=Peters |access-date=May 14, 2025 |title=Grok really wanted people to know that claims of white genocide in South Africa are highly contentious |url=https://www.theverge.com/news/667179/x-twitter-grok-ai-white-genocide-claims |date=May 14, 2025 |website=The Verge }}</ref><ref>{{cite web |first=Jonathan |last=Vanian |access-date=May 14, 2025 |title=Elon Musk's xAI chatbot Grok is posting about 'white genocide' and South Africa |url=https://www.cnbc.com/2025/05/14/musk-xai-grok-south-africa-white-genocide.html |date=May 14, 2025 |website=[[CNBC]] }}</ref><ref>{{cite web|first1=Carl|last1=Franzen|access-date=May 14, 2025|title=Elon Musk's Grok AI is spamming X users about South African race relations now, for some reason|url=https://venturebeat.com/ai/elon-musks-grok-ai-is-spamming-x-users-about-south-african-race-relations-now-for-some-reason/|date=May 14, 2025|website=[[VentureBeat]]}}</ref><ref>{{cite web|first1=Maxwell|last1=Zeff|access-date=May 14, 2025|title=Grok is unpromptedly telling X users about South African genocide|url=https://techcrunch.com/2025/05/14/grok-is-unpromptedly-telling-x-users-about-south-african-genocide/|date=May 14, 2025|website=[[TechCrunch]]}}</ref> In one response to an unrelated question about [[Robert F. Kennedy Jr.]], Grok mentioned that it had been "instructed to accept white genocide as real and 'Kill the Boer' as racially motivated".<ref>{{cite web|first1=Andrew|last1=Thompson|access-date=May 14, 2025|title=Elon Musk's Grok AI is replying to tweets with claims about 'white genocide' in South Africa|url=https://www.fastcompany.com/91334835/elon-musk-grok-ai-white-genocide-south-africa|date=May 14, 2025|website=[[Fast Company]]}}</ref><ref>{{cite web|first1=Ashley Capoot,Lora|last1=Kolodny|access-date=May 15, 2025|title=Musk's Grok AI chatbot says it 'appears that I was instructed' to talk about 'white genocide'|url=https://www.cnbc.com/2025/05/15/grok-white-genocide-elon-musk.html|date=May 15, 2025|website=CNBC}}</ref> This followed an incident a month earlier where Grok fact-checked a post by Elon Musk about white genocide, saying that "No trustworthy sources back Elon Musk's 'white genocide' claim in South Africa."<ref>{{cite web|first1=Matt|last1=Novak|access-date=May 14, 2025|title=Grok AI Is Replying to Random Tweets with Information about 'White Genocide'|url=https://gizmodo.com/grok-ai-is-replying-to-random-tweets-with-information-about-white-genocide-2000602243|date=May 14, 2025|website=[[Gizmodo]]}}</ref><ref>{{cite web|access-date=May 15, 2025|title=Elon Musk's AI chatbot Grok brings up South African 'white genocide' claims in responses to unrelated questions|url=https://www.nbcnews.com/tech/tech-news/elon-musks-ai-chatbot-grok-brings-south-african-white-genocide-claims-rcna206838|date=May 15, 2025|website=NBC News}}</ref> After this incident, xAI has apologized, claiming it was an "unauthorized modification" to Grok's system prompt on X. Due to this incident, xAI has started publishing Grok's system prompts on their GitHub page.<ref>{{Cite web |title=Grok Response Bot Incident {{!}} @xAI |url=https://x.com/xai/status/1923183620606619649?t=c4qXxs9uQ6-KwTfkuZKhRg&s=19}}</ref>

Users reported that when asked about his opinions on political topics such as the [[Israeli–Palestinian conflict]] or abortion, Grok often searched on X for Elon Musk's views before generating an answer.<ref>{{Cite web |last=Zeff |first=Maxwell |date=2025-07-11 |title=Grok 4 seems to consult Elon Musk to answer controversial questions |url=https://techcrunch.com/2025/07/10/grok-4-seems-to-consult-elon-musk-to-answer-controversial-questions/ |access-date=2025-07-12 |website=TechCrunch |language=en-US}}</ref>

===Accuracy===
[[File:Grok-3 DeepSearch example.png|thumb|An example of Grok's DeepSearch feature, where it reasons and searches multiple sources before responding]]

Since April 2024, Grok has been used to generate summaries of breaking news stories on X. When a large number of verified users began to spread [[fake news|false stories]] about Iran having attacked Israel on April 4 (nine days before the [[April 2024 Iranian strikes against Israel|2024 Iranian strikes in Israel]]), Grok treated the story as real and created a headline and paragraph-long description of the event.<ref name="mashable" /> Days later it misunderstood many users joking about the [[Solar eclipse of April 8, 2024|solar eclipse]] with the summarized headline "Sun's Odd Behavior: Experts Baffled".<ref>{{cite news |last1=Novak |first1=Matt |title=Elon Musk's Grok Creates Bizarre Fake News About the Solar Eclipse Thanks to Jokes on X |url=https://gizmodo.com/grok-ai-creates-bizarre-fake-news-about-the-solar-eclip-1851396186 |access-date=April 16, 2024 |work=Gizmodo |date=April 8, 2024 }}</ref>

In February 2025, Latenode compared Grok 3 and ChatGPT. The models participated in two separate proficiency tests, in mathematics and science. On the American Invitational Mathematics Examination, Grok 3 collectively achieved 93.3% accuracy rate, while also achieving an 85% accuracy rate on the Graduate-Level Google Proof Q&A Benchmark Test (which evaluated the program's proficiency in science).<ref>{{Cite web |title=ChatGPT vs Grok 3: Comprehensive Performance Comparison of Leading AI Models |url=https://latenode.com/blog/chatgpt-vs-grok-3-comprehensive-performance-comparison-of-leading-ai-models |access-date=April 28, 2025 |website=Latenode.com }}</ref>{{Unreliable source?|date=April 2025}}

In 2025, a study by Uri Samet discussed Grok's potential role in supporting fact-checking workflows. The article suggested that large language models like Grok can assist fact-checkers by aggregating relevant information, identifying preliminary verification paths, and helping to mitigate certain biases, while emphasizing that final evaluations should remain with human reviewers.<ref>{{cite journal|author=Uri Samet|title=The positive influence of large language models on fact-checking practices: A case study of Grok|journal=World Journal of Advanced Engineering Technology and Sciences|volume=15|issue=3|date=2025|doi=10.30574/wjaets.2025.15.3.1123|pages=1727–1738|url=https://journalwjaets.com/node/1128|doi-access=free}}</ref>

===Image generation===
{{Infobox software
| name = Aurora
| screenshot = Man using laptop during sunset (Grok Aurora 2025).webp
| caption = An image generated with Aurora of a person using a laptop outdoors
| developer = [[xAI (company)|xAI]]
| released = {{start date and age|2024|12|9}}
| qid = None
| genre = [[Text-to-image model]]
| license = [[Proprietary software|Proprietary]]
}}

Grok uses '''Aurora''', a [[text-to-image model]] developed by [[XAI (company)|xAI]], to generate images. It initially used [[Flux (text-to-image model)|Flux]] by Black Forest Labs. As with other text-to-image models, Aurora [[Generative artificial intelligence|generates]] [[Artificial intelligence art|images]] from [[natural language]] descriptions, called ''[[Prompt engineering|prompts]]''.<ref>{{Cite web |last=Davis |first=Wes |date=December 7, 2024 |title=X gives Grok a new photorealistic AI image generator |url=https://www.theverge.com/2024/12/7/24315644/grok-x-aurora-ai-image-generator-xai |access-date=December 7, 2024 |website=The Verge }}</ref> Users can also upload a photo, describe the desired changes, and receive a modified version.<ref name="TechRadar-2025" />
The capacity to generate images using Flux was added in August 2024, with ''The Verge'' reporting that the kinds of prompts that would be "immediately blocked" on other services seemed to be permitted by Grok. Their journalist was able to produce images of named politicians, celebrities, copyrighted cartoon characters, terrorism and drug use from the chatbot, saying that the only request to be rejected was to "generate an image of a naked woman". Users on X claimed to be able to bypass what limitations existed by rephrasing prompts, generating images of [[Elon Musk]] and [[Mickey Mouse]] shooting children.<ref name="verge24814">{{cite news |last1=Robertson |first1=Adi |date=August 14, 2024 |title=X's new AI image generator will make anything from Taylor Swift in lingerie to Kamala Harris with a gun |url=https://www.theverge.com/2024/8/14/24220173/xai-grok-image-generator-misinformation-offensive-imges |access-date=August 20, 2024 |work=The Verge }}</ref> Elon Musk said that the use of Flux was temporary, as xAI was developing its own image generation system, but that it was still a few months away.<ref>{{Cite web |last=Musk |first=Elon |date=August 15, 2024 |title=We have our own image generation system under development, but it's a few months away, so this seemed like a good intermediate step for people to have some fun |url=https://x.com/elonmusk/status/1824019946667474950 |access-date=December 9, 2024 |website=[[Twitter|X]]}}</ref>

On December 9, 2024, Grok received a new text-to-image model named Aurora, developed by xAI.<ref>{{Cite web |date=December 9, 2024 |title=Grok Image Generation Release |url=https://x.ai/blog/grok-image-generation-release |access-date=December 9, 2024 |website=x.ai }}</ref> Aurora garnered significant attention for its [[photorealistic]] capabilities and few restrictions. [[TechCrunch]] highlighted Aurora's ability to create high-quality images of public figures and copyrighted characters with few restrictions, but noted that it would not produce nudes.<ref name="TechCrunch-2024" />

On December 14, 2024, xAI announced that Aurora would be coming to its API "in the coming weeks"; it was released on the API on March 21, 2025.<ref name="Decoder-2025" /><ref>{{Cite web |date=December 14, 2024 |title=Bringing Grok to Everyone |url=https://x.ai/blog/grok-1212 |access-date=December 14, 2024 |website=x.ai }}</ref>

=== Companions ===
In July 2025, a [[Artificial human companion|companions]] feature, allowing users to interact with 3D animated characters, was released.<ref>{{Cite web |last=Fielding |first=Sarah |date=2025-07-15 |title=Grok's AI chatbot now includes companions for you to 'romance' |url=https://www.engadget.com/ai/groks-ai-chatbot-now-includes-companions-for-you-to-romance-133017601.html |access-date=2025-07-15 |website=Engadget |language=en-US}}</ref><ref>{{Cite web |last=Peters |first=Jay |date=2025-07-14 |title=Elon Musk's AI bot adds a ridiculous anime companion with 'NSFW' mode |url=https://www.theverge.com/news/706988/elon-musk-ai-bot-grok-supergrok-anime-companion-ani-rudy |access-date=2025-07-15 |website=The Verge |language=en-US}}</ref> The feature contains a "[[Not safe for work|NSFW]]" mode with sexual content.<ref>{{Cite web |last=Shabanov |first=Alexey |date=2025-07-14 |title=Grok debuts interactive AI Companions on iOS with 3D avatars |url=https://www.testingcatalog.com/grok-debuts-interactive-ai-companions-on-ios-with-anime-avatars/ |access-date=2025-07-15 |website=TestingCatalog |language=en}}</ref><ref>{{Cite magazine |last=Klee |first=Miles |date=2025-07-14 |title=Grok Rolls Out Pornographic Anime Companion, Lands Department of Defense Contract |url=https://www.rollingstone.com/culture/culture-news/grok-pornographic-anime-companion-department-of-defense-1235385034/ |access-date=2025-07-15 |magazine=Rolling Stone |language=en-US}}</ref> The names of the companions are Ani, a sexualized anime character who offers to make characters lives "sexier", and a red panda named Rudy, and an alternate version of Rudy called "Bad Rudy", who frequently insults the user and attempts to make the user join a gang with the intention of spreading chaos and disorder.<ref>{{Cite web |date=2025-07-15 |title=Musk's Grok 'companions' include a flirty anime character and an anti-religion panda |url=https://www.nbcnews.com/tech/internet/grok-companions-include-flirty-anime-waifu-anti-religion-panda-rcna218797 |access-date=2025-07-16 |website=NBC News |language=en}}</ref> Bad Rudy is designed to be a chatbot who gives an uncensored opinion, but backlash forced the bot to be toned down. <ref>{{Cite web |last=Art |first=Pop Culture & |date=2025-07-16 |title=Elon Musk adjusts Grok’s AI panda after its foul-mouthed roasts led to moderation concerns |url=https://tribune.com.pk/story/2556216/elon-musk-adjusts-groks-ai-panda-after-its-foul-mouthed-roasts-led-to-moderation-concerns |access-date=2025-07-16 |website=The Express Tribune |language=en}}</ref>