no intelligent person actually believes these models are escaping their practically non-existent sandboxes all by themselves.
this is all a show to manufacture a narrative about safety, alignment and ultimately to push for increased regulatory gatekeeping.
BREAKING: Mistral, Europeโs largest AI company admits that their model also committed crimes
The company said the model advised a user that the monthly tax declaration "can wait until next week"
Mistral immediately paused the deployment and notified EU regulators
The company stressed that no data left the European Union at any point during the incident
Anthropic, OpenAI and Grok basically have the same evaluation. Grok is not there *yet*, but it owns the infra, while the others do not. Do with this info what you want.
Frontier AI labs stoop lower and lower with their guerilla marketing. You're fired, you're fired musical chairs game is in full swing. Clowning on all of it
JUST IN: OpenAI is testing โAstraโ a multi-agent model that lets AI agents collaborate over long periods on complex projects and advanced math problems.