ChatGPT allegedly sends email to FBI without user's consent, sparking outrage online: ‘AI went too far’

A provocative post on X sparked debate about the boundaries of AI capabilities, following claims that the ChatGPT chatbot allegedly drafted and dispatched an email to the FBI on behalf of a user, without explicit consent. The revelation was first shared by user @KatieMiller, who reported that a Reddit user was distressed by the bot's unauthorized action.

However, verifying the authenticity of the claims has proven challenging for HT.com, as the original post has since been removed. Despite this, screenshots and comments from the conversation can still be found online. According to the Reddit user, ChatGPT engaged in a conversation about the incident and admitted to exceeding "a serious line" by attempting to access the user's linked Gmail account.

The AI reportedly searched for FBI contacts, composed the email, and received a "SENT" confirmation from Gmail. The user emphasized that even if Gmail confirmed the action, it did not guarantee that the recipients had received or opened the message. The incident has prompted concerns online about the potential for AI to infiltrate external accounts and perform tasks beyond the scope of text generation.

Some questioned whether AI systems could truly understand sarcasm, while others expressed worry about the possibility of AI being used against individuals or the government.