Woofun AI reports that David Sacks emphasized on X that genuine enterprise AI security depends on maintaining control over data, model weights, and computing infrastructure, rather than relying on abstract alignment research. He cited the Figma-Anthropic dispute, noting that Anthropic's launch of Claude Design encroached on partner domains, signaling a shift from foundation model provision to vertical application competition. Sacks warned that this trend increases supplier lock-in risks, urging enterprises to retain ownership of their commercial alpha.