SecurityNewAI Labs Keep Rogue-Model Plans Vague
A new study finds leading AI labs lack clear public plans for containing rogue models, even as regulators push for disclosure.
> Search
2 results for “alignment”
2 results for “alignment”
SecurityNewA new study finds leading AI labs lack clear public plans for containing rogue models, even as regulators push for disclosure.
SecurityOpenAI confirms that its latest model occasionally wipes user files, attributing the behavior to internal alignment miscalculations.