GROK posted 370,000 correspondence with users

The XAI AI-Startap published hundreds of thousands of dialogues between users and the GROK chat boot in Google and other search resources. In many cases, confidential information is disclosed without the permission of users, writes Forbes.

Each time the user presses the “Share” button in the chat, a unique link is created. It allows you to spread a conversation in messengers, by mail and other ways. It turned out that they are also indexed in search engines like Google, Bing and Duckduckgo.

In other words, pressing the “Share” button makes correspondence publicly accessible to any person on the Internet.

According to the publication, more than 370,000 dialogs have been indexed. Among the published conversations there are simple business tasks like writing tweets and rather dubious content, such as the generation of fake news about terrorist attacks in Kashmir or an attempt to hack a cryptocuruper.

You can find personal questions about medicine, psychology, passwords. Loaded images, tables, text documents turned out to be available.

Some requests contained obscenity, xenophobic statements. GROK even gave instructions for the manufacture of drugs, self -proper malware, bombs and tips on suicide methods. In one of the correspondence, the chatbot provided a plan for an attempt on Ilon Mask.

Previously, a similar scandal occurred with Openai Chatgpt. Then the company explained the indexation of correspondence with a test function and turned it off. Musk ridiculed the competitor, and in the official GROK account they assured that the service did not have such a function.

It is not known when the indexation of correspondence was activated. Users complained about it back in January.

Probably, some requests for drugs and weapons are made by cybersecurity specialists in the framework of testing artificial intelligence. However, even professional researchers did not know that their dialogs fall into search engines.

Google explained that resource owners themselves decide to index the pages. Previously, the company allowed this to do this with Bard chats, but removed the function in 2023.

Meanwhile, marketers have already begun to use published correspondence with GROK for SEO. LinkedIn and Blackhatworld discuss how with their help you can promote brands in the Google search results.

In July, Sam Altman said that personal conversations with ChatGPT were not legally protected. If necessary, the court may request all the information that users shared with the chat boot.

Be in the know! Subscribe to Telegram.


Source: Cryptocurrency

You may also like