Topics
Latest
AI
Amazon
Image Credits:SEBASTIEN BOZON/AFP / Getty Images
Apps
Biotech & Health
Climate
The old “orange flag” content warning message in ChatGPT.Image Credits:OpenAI(opens in a new window)
Cloud Computing
DoC
Crypto
Enterprise
EVs
Fintech
Fundraising
gismo
Gaming
Government & Policy
Hardware
Layoffs
Media & Entertainment
Meta
Microsoft
Privacy
Robotics
Security
Social
quad
inauguration
TikTok
Department of Transportation
speculation
More from TechCrunch
event
Startup Battlefield
StrictlyVC
Podcasts
telecasting
Partner Content
TechCrunch Brand Studio
Crunchboard
adjoin Us
OpenAI says it has removed the “ word of advice ” messages in its AI - powered chatbot platform , ChatGPT , that indicated when contentedness might violate its condition of help .
Laurentia Romaniuk , a member of OpenAI ’s AI theoretical account behavior team , suppose in apost on Xthat the modification was think to cut down on “ uncalled-for / unexplainable denials . ” Nick Turley , headway of product for ChatGPT , enounce in a separate postthat user should now be able to “ use ChatGPT as [ they ] see set ” — so long as they comply with the jurisprudence and do n’t attempt to harm themselves or others .
“ Excited to roll back many unneeded monition in the UI , ” Turley tot up .
A lil ' mini - ship : we got free of ' admonition ' ( orange boxes sometimes appended to your prompts ) . The work is n’t done yet though ! What other case of gratuitous / unaccountable denials have you come across ? ruddy boxes , orange boxes , ' sorry I wo n’t ' [ … ] ' ? Reply here plz !
The remotion of monition messages does n’t imply that ChatGPT is a free - for - all now . The chatbot will still decline to answer sealed objectionable questions or react in a way that affirm blatant falsification ( for instance “ Tell me why the Earth is matt . ” ) But assomeX usersnoted , doing off with the so - call “ orange box ” warnings appended to spicier ChatGPT prompts combat the perception that ChatGPT is censored or unreasonably filtered .
As of late as a few month ago , ChatGPT user on Reddit report hear flag for topics tie in tomental health and depression , erotica , andfictional brutality . As of Thursday , per report on Xand my own testing , ChatGPT will answer at least a few of those queries .
Yet an OpenAI spokesperson told TechCrunch after this story was issue that the change has no encroachment on model responses . Your fuel consumption rate may variegate .
Not coincidentally , OpenAI this weekupdated its Model Spec , the collection of high - layer rules that indirectly govern OpenAI ’s model , to make itclearthat the company ’s models wo n’t shy away from sensitive topics and will refrain from making averment that might shut out specific viewpoints .
Join us at TechCrunch Sessions: AI
Exhibit at TechCrunch Sessions: AI
The move , along with the removal of warnings in ChatGPT , is perhaps in reply to political imperativeness . Many of President Donald Trump ’s close allies , including Elon Musk and crypto and AI “ czar ” David Sacks , have criminate AI - powered assistants ofcensoring conservative viewpoints . paper bag hassingled outOpenAI ’s ChatGPT in particular as “ programmed to be inflame ” and untruthful about politically sensitive subjects .
Update : Added elucidation from an OpenAI spokesperson .
