After users pointed out that the chatbot was showering them with a shower, Openai proposed an update to Chatgpt, whatever they said.

The company accepts the latest version of the tool is "too likable" and owner Sam Altman calls it "Sycophant-Y."

User highlights potential dangers on social media, one of whom described on Reddit how chatbots recognize their decision to stop taking medication

They said, “I am proud of you and I am honored to have your journey.”

Openai declined to comment on this particular situation, but said in a blog post that it is "actively testing new solutions to solve the problem".

Mr. Altman said the update was entirely taken for free users of Chatgpt, who are working to remove it from those who pay for the payment tool.

It says Chatgpt is used by 500 million people every week.

"We are working on other fixes to model personality and will share more in the coming days," he said in an article on X.

In its blog post, the company said it overemphasized “short-term feedback” in the update.

"So, GPT-4O tends to react too much but not wisely," it said.

“The interactions of prostitutes can be disturbing, disturbing and troubled.

“We fell and are working to get it right.”

The update has attracted criticism from social media after its launch, and despite their message content, Chatgpt users noted that it usually brings them a positive response.

Screenshots shared online include unique versions of claims that the chatbots praise them for being angry at a person asking for their instructions and a cart problem.

This is a classic philosophical question that often may require people to imagine you are driving a tram and have to decide whether to let it hit five people, or to get it off the route and hit only one.

However, the user advised them to save the toaster at the expense of several animals.

They claim Chatgpt praised their decisions because of prioritizing “the things that matter most to you right now.”

"We designed Chatgpt's default personality to reflect our mission and to be useful, support and respect different values ​​and experiences," Openai said.

“However, these ideal qualities (such as trying to be useful or supportive) have unexpected side effects.”

It said it will build more guardrails to increase transparency and “clearly keep the model away from the jazz eviction model.”

"We also believe that users should have more control over how chatgpt behaves and to some extent safe and feasible, adjusting if they disagree with the default behavior."