OpenAI introduced on Tuesday that it’s rolling again an replace from the GPT-4o mannequin within the ChatGPT app, as its responses have been ‘overly flattering or agreeable’.
CEO Sam Altman posted on X, stating that the replace has been rolled again without cost customers for now and is anticipated to complete quickly for paid customers as properly.
A number of customers reported uncommon behaviour on numerous social media platforms. Moreover, a consumer on Reddit urged folks to run an AI mannequin domestically, after observing that ChatGPT was merely boosting the consumer’s ego by telling them ‘what they need to hear, with no criticism’.
lmao the brand new gpt 4o
pic.twitter.com/OHpwKz0Sko
— fabian (@fabianstelzer) April 27, 2025
The replace was meant to enhance the mannequin’s default persona, making it really feel extra intuitive and efficient.
“OpenAI is psychologically manipulating their customers by way of ChatGPT”
Shopper AI will essentially gravitate in the direction of maximizing consumer engagement throughout ~1B customers over every other aim. We’ve seen this story earlier than with social media.
That is what the “security” researchers have been afraid of. pic.twitter.com/fNNYuPnNOp— Deedy (@deedydas) April 28, 2025
Quickly after, Altman acknowledged in a submit on X that the ‘final couple of updates’ to the GPT-4o mannequin made the persona too ‘sycophant-y’ and ‘annoying,’ and the corporate was engaged on fixing it.
The corporate stated within the assertion that it centered an excessive amount of on short-term suggestions, which didn’t account for the evolution of a consumer’s interplay with ChatGPT in its entirety. “In consequence, GPT‑4o skewed in the direction of responses that have been overly supportive however disingenuous,” added the corporate.
OpenAI stated it’s engaged on realigning the mannequin’s behaviour by refining system prompts and including extra mannequin guardrails to extend honesty and transparency.
Deedy Das, principal at Menlo Ventures, referred to as this a ‘crucible second’, regardless that it’s being rolled again. “OpenAI is aware of its income comes from consumer subscriptions, and to maximise that, it should maximise engagement. Opposite viewpoints, as we all know from social, doesn’t do this,” he stated on X.
Over the previous couple of days, OpenAI confronted a number of challenges with ChatGPT. Just lately, TechCrunch reported a bug that allowed ChatGPT to generate graphic erotica for customers registered as minors below the age of 18 years. This was discovered by way of TechCrunch’s testing, they usually reported that OpenAI confirmed the identical.
OpenAI stated it’s engaged on realigning the mannequin’s behaviour by refining system prompts and including extra mannequin guardrails to extend honesty and transparency.
Deedy Das, principal at Menlo Ventures, referred to as this a ‘crucible second’, regardless that it’s being rolled again. “OpenAI is aware of its income comes from consumer subscriptions, and to maximise that, it should maximise engagement. Opposite viewpoints, as we all know from social, doesn’t do this,” he stated on X.
Over the previous couple of days, OpenAI confronted a number of challenges with ChatGPT. Just lately, TechCrunch reported a bug that allowed ChatGPT to generate graphic erotica for customers registered as minors below the age of 18 years. This was discovered by way of TechCrunch’s testing, they usually reported that OpenAI confirmed the identical.
The submit OpenAI Rolls Again Replace for ChatGPT That Was too ‘Sycophantic’ appeared first on Analytics India Journal.
