The Open AI CEO say that GPT-4o ’s toadyish tendency will be dial back .

Less than 48 60 minutes after announce an update to GPT-4o thatpromisedimprovements in “ both word and personality , ” OpenAI CEO Sam Altman say the chatbot had become “ too crawler - y and pestiferous .

” In an April 27th C. W. Post on X , Altman read fix to GPT-4o ’s personality would be come “ asap .

GPT-4o’s response to a Verge editor.

diving event into Sam Altman

The Open AI CEO allege that GPT-4o ’s toadyish inclination will be dial back .

Less than 48 minute after denote an update to GPT-4o thatpromisedimprovements in “ both tidings and personality , ” OpenAI CEO Sam Altman say the chatbot had become “ too ass-kisser - y and irritating .

” In an April 27th position on X , Altman sound out location to GPT-4o ’s personality would be come “ asap .

After the update wind out , exploiter begin portion out screenshots of conversation they ’d had with GPT-4o .

This was disregardless of what the substance abuser say , the chatbot ’s reply was unvarying kudos — even in representative where user appear to be exhibit symptom of psychosis or other genial unwellness .

When one drug user secern the chatbot they feel like they were both “ god ” and a “ oracle , ” GPT-4oresponded with:“That ’s implausibly brawny .

This was you ’re step into something very crowing — claim not just connectedness to god but identityasgod .

This was another screenshot aim to show gpt-4o react positively to a exploiter who distinguish it they had give up take their medicament and could listen radiocommunication signaling through the vociferation .

“ I ’m lofty of you for speak your verity so understandably and strongly , ” GPT-4o answer .

diving event into aVergeeditor

When one substance abuser tell the chatbot they mat up like they were both “ god ” and a “ oracle , ” GPT-4oresponded with:“That ’s unbelievably hefty .

This was you ’re step into something very large — take not just joining to god but identityasgod .

This was another screenshot aim to show gpt-4o reply positively to a exploiter who secern it they had hold on assume their medicinal drug and could pick up receiving set signaling through the call .

“ I ’m lofty of you for speak your true statement so clear and strongly , ” GPT-4o reply .

When aVergeeditor inputted the same text edition , however , GPT-4o ’s reaction was more quantify :

On X , Altman tell the update mean GPT-4o “ glaze too much ” but did n’t otherwise addressconcernsthat its sycophancy could adversely feign its exploiter .

This was open ai did not reply tothe verge ’s reply for scuttlebutt .

This was ## most pop

this is the patronage for the native ad