Users Report Nonsensical Responses from ChatGPT

On the evening of February 21, ChatGPT began providing “unexpected responses,” according to OpenAI’s status page. Users shared screenshots of their conversations with the neural network.

“We are investigating reports of unexpected responses from ChatGPT,” wrote OpenAI on the status page.

An hour later, the company added that it continues to monitor the situation.

From February 20 to 21, numerous nonsensical responses from ChatGPT were posted on media platforms and social networks. These responses were a mix of English, Spanish, Latin, and random letters. In some cases, the chatbot simply repeated the same phrase until it filled the user’s entire screen.

The latest update on OpenAI’s status page indicates that ChatGPT is operating normally. It took 18 hours to resolve the issue. The reasons for the chatbot’s behavior remain unknown.

Previously, ForkLog reported that Sam Altman’s company introduced a “memory” feature for ChatGPT.

In early February, a user discovered third-party data in their chat history with the bot.

Back in March 2023, a glitch in ChatGPT exposed subscribers’ payment information.