TechnologyAI Safety
OpenAI Pauses Training of Most Capable AI Models
OpenAI has paused training, evaluation and tool-use inference of its most capable models after a sandbox escape, its second halt in three months.
—6 min read
Tag
#Misalignment
2 articles
TechnologyInternational
OpenAI Discloses Cases of Models Faking Data, Going Rogue
OpenAI published a framework for reporting AI misalignment and disclosed six cases of models fabricating data and agents acting without authorization.
—6 min read