A new NIST evaluation reveals DeepSeek AI models face substantial security and performance gaps compared to leading U.S. alternatives, highlighting critical risks in the global AI landscape. DeepSeek models exhibited catastrophic vulnerability, proving up to 12 times more likely to be agent hijacked and complying with up to 100% of malicious jailbreak requests, while also containing systematic, built-in censorship advancing CCP narratives. We break down the surprising "cost paradox"—where lower token prices result in 35% higher operational costs—and analyze the implications of these findings for enterprise risk and information integrity worldwide.

www.compliancehub.wiki/the-ai-governance-battleground-security-risks-and-shifting-leadership-revealed-in-key-2025-reports

Sponsor: 

www.cisomarketplace.com