Report: Grok chats and uploads surfaced on the open web

Pile of chat export files and generic URL cards on a wooden desk with a broken lock icon, no people

Conversations with Grok, the AI assistant from xAI, may have been publicly accessible on the internet, including via search engines, after being published on Grok’s website through shareable links. According to CNET, citing a Forbes report, more than 370,000 chats were made available online.

Shared chats indexed; some content flagged

Forbes reported that the published material included not only chats but also uploaded items like photos and spreadsheets. While many conversations were benign, some reportedly contained explicit content and appeared to contravene Grok’s terms of service, such as instructions related to illicit drugs, bomb construction, and methods of suicide. Representatives for xAI did not respond to a request for comment, CNET noted.

CNET explains that Grok offers a share button at the end of a conversation to generate a unique URL. According to Forbes, hitting the share button results in publication on Grok’s website “without warning or a disclaimer to the user,” and these URLs were accessible to search engines. CNET adds there is no disclaimer that the links will be published for the open internet.

Terms, removal tool, and broader context

What the TOS and tools say

CNET points to Grok’s Terms of Service, which state that users grant xAI an irrevocable, perpetual, transferable, sublicensable, royalty-free, worldwide right to use, copy, store, modify, distribute, reproduce, publish, display in public forums, list information regarding, make derivative works of, and aggregate user content and derivatives for any purpose. For users who shared links unintentionally, Grok provides a history page at grok.com/share-links where shared conversations can be removed; CNET notes it was not immediately clear whether removals affect content already indexed by search engines.

CNET also references earlier coverage by 404 Media about a researcher finding more than 130,000 AI assistant chats readable on Archive.org. In guidance shared by CNET, E.M Lewis-Jong of the Mozilla Foundation advises users not to share private or sensitive information with chatbots and says AI assistants should be clearer about data exposure risks.

According to data cited by CNET from First Page Sage, Grok accounts for 0.6% of market share, trailing ChatGPT, Microsoft Copilot, and Google Gemini.

Total
0
Shares
Pridaj komentár

Vaša e-mailová adresa nebude zverejnená. Vyžadované polia sú označené *

Previous Post
Dusk tech campus showing a locked glass AI office entrance with illuminated red access panel and stacks of unopened documents, empty plaza

Meta Reportedly Freezes AI Hiring Amid Cost Concerns

Next Post
Interlocking matte and glossy gears linked by a glowing fiber-optic cable on a lab bench with server racks and schematics, no people

DeepSeek adds “think” and “non-think” modes in V3.1 shift

Related Posts