Physical Address
Indirizzo: Via Mario Greco 60, Buttigliera Alta, 10090, Torino, Italy
The office of Ankara’s chief prosecutor launched a formal investigation on Wednesday into Grok, the artificial intelligence chatbot developed by the Elon Musk-founded company xAI, amid a flurry of offensive responses it sent to many users.
Grok, available on the social media platform X, was reported to have produced several posts containing offensive language and even swearing, according to the posts shared by the users late on Tuesday.
The chatbot, integrated into Musk’s X, formerly Twitter, also reportedly generated offensive content about President Recep Tayyip Erdoğan when asked certain questions in Turkish, media said.
According to information obtained by Anadolu Agency (AA), the probe was launched due to insulting words in the responses given to many users, for the crimes including “publicly insulting religious values adopted by a segment of the public,” “insulting the President” and “violating Law No. 5816 on Crimes Committed Against Atatürk.”
Issues of political bias, hate speech and accuracy of AI chatbots have been a concern since at least the launch of OpenAI’s ChatGPT in 2022. Grok, meanwhile, was also dropping content accused of anti-Semitic tropes and praise for Adolf Hitler.
Earlier this week, it was forced to withdraw some posts, which were deemed “inappropriate” after complaints from X users and the Anti-Defamation League (ADL).
“We are aware of recent posts made by Grok and are actively working to remove the inappropriate posts,” Grok posted on X, related to the incident.
The posts shared on X in Türkiye also stirred debate, confusion and concerns over the provided answers among users.
Neither X nor its owner, Musk, has commented in relation to the case so far.
This also marked the first official investigation into any chatbot in Türkiye since the launch of ChatGPT. Chatbots, as computer programs, are designed to stimulate real-time conversation with humans, and have grown increasingly popular, but are also known to produce incorrect data and information, often dubbed as “hallucinations.”
Last month, Musk promised an upgrade to Grok, suggesting there was “far too much garbage in any foundation model trained on uncorrected data.”
“We have improved Grok significantly. You should notice a difference when you ask Grok questions,” Musk said in a recent post on X.
Turkish media reports indicated on Wednesday that the decision to block access to the chatbot was sent to the Information and Communication Technologies Authority (BTK).
Although reports pointed out Grok was blocked in the country, when checked on both X and its website, the chatbot was still responding to queries as of 11 a.m. local time (8 a.m. GMT).
