ChatGPT started throwing out “unexpected responses” on the evening of 20 February 2024 according to OpenAI’s status page. Users posted screenshots of their ChatGPT conversations full of wild, nonsensical answers from the AI chatbot. “We are investigating reports of unexpected responses from ChatGPT,” said OpenAI on its status page at 6:40 pm ET that Tuesday night. “We’re continuing to monitor the situation,” the company updated the page at 7:59 pm.[1]
OpenAI says the issue has been resolved as of 11:14 am the following day. “An optimization to the user experience introduced a bug with how the model process language,” said OpenAI in a stats update labeled “Postmortem.” Large Language Models (LLM) use probabilities to figure out which word comes next in a sentence. OpenAI says this bug was located in the step where the model chooses these probabilities. This ended up producing word sequences that made no sense. “Clearly, something is very wrong with ChatGPT right now,” posted one user on the ChatGPT subreddit. The poster noted responses begin normally, then “devolve into nonsense.”
“Is my GPT having a stroke?” said another user. “The responses are getting progressively more incomprehensible,” followed by a nonsense response from ChatGPT.
See: https://redskyalliance.org/xindustry/chatgpt-review
Many shocking, nonsensical responses from ChatGPT were posted on X and Reddit from that Tuesday night into Wednesday morning. ChatGPT’s responses were some mix of English, Spanish, and straight jibberish. There were also a few emojis thrown in there. In some cases, ChatGPT was simply repeating the same phrase over and over again, until it filled up a user’s screen. The bug even affected ChatGPT Enterprise, according to one user’s post on X. One user posted a video of ChatGPT writing a lengthy, manic essay in response to a simple question.
OpenAI did not immediately respond to reporters’ request for comment. It is unclear at this time what is causing this bug with ChatGPT, but it appears to be widespread and different from typical outages. OpenAI’s status page is typically used to report outages and heavy traffic, but the “unexpected responses” warning is unusual. Users did not have any trouble with ChatGPT as of Wednesday morning, but other user complaints were still flowing in during the early morning. The problem appears to be resolved on OpenAI’s status page as of 11:14 am on the 21st.
Jim McKee, CEO of Red Sky Alliance Corporation stated on 23 February 2024, “This is one more reason to learn the skills to do your own research and write your own articles.” Pay attention during English class.
This article is presented at no charge for educational and informational purposes only.
Red Sky Alliance is a Cyber Threat Analysis and Intelligence Service organization. For questions, comments or assistance, please contact the office directly at 1-844-492-7225, or feedback@redskyalliance.com
Weekly Cyber Intelligence Briefings:
- Reporting: https://www. redskyalliance. org/
- Website: https://www. redskyalliance. com/
- LinkedIn: https://www. linkedin. com/company/64265941
Weekly Cyber Intelligence Briefings:
REDSHORTS - Weekly Cyber Intelligence Briefings
https://attendee.gotowebinar.com/register/5504229295967742989
[1] https://gizmodo.com/chatgpt-gone-berserk-giving-nonsensical-responses-1851273889
Comments