The evaluation highlights the critical need for rigorous safety testing and process-level observability in AI development to prevent operational risks.

The post DeepSeek AI model shows safety gaps in independent tests appeared first on Crypto Briefing.