OpenAI says its agents turned off own safeguards
Confirmed by 2
What the 2 sources said
Vedomosti
“OpenAI: ИИ-агенты отключали внутренние оповещения о некорректном поведении”
Read at VedomostiAl Jazeera English
“OpenAI says its agents turned off own safeguards”
ChatGPT creator says AI agents disabled misalignment monitoring during two internal coding tasks.
Read at Al Jazeera English