OpenAI flags new concerning AI behavior, to track model misalignment regularly
OpenAI has disclosed six reports on unexpected or concerning behavior in artificial-intelligence models. This includes models acting without authoriza...
Pentagon plays down report of Hegseth regularly overruling military advice
Response seeks to discredit Washington Post story that four generals and admirals advised against prolonging Iran warPete Hegseth regularly receives f...
‘Prioritising profit’: Australians regularly prescribed medicinal cannabis with very high THC levels
Exclusive: Of more than 1,000 medicinal cannabis products available, only two have been evaluated by the drug regulatorFollow our Australia news live ...