**[David Sacks: True Enterprise AI Security is "Control," Not Abstract Alignment Research]** Odaily Planet Daily reported that David Sacks commented on an interview with Palantir CEO Alex Karp via the X platform, stating that some traditional media interpreted Karp's remarks as "emotional expression." However, Sacks argued that Karp's perspective actually highlights the core issue of enterprise-level AI security. Sacks emphasized that true enterprise AI security is not about abstract "AI alignment" research or government-style regulatory frameworks. Instead, it lies in a company's complete control over its own data, model weights, and computational infrastructure to prevent core intellectual assets from being absorbed by model vendors and turned into competitive advantages for their own products. He cited Karp's viewpoint, noting that enterprise clients are genuinely concerned about control over computing resources, models, and data stacks—essentially ensuring that "ownership of production materials is not transferred." Sacks also referenced the controversy surrounding the collaboration between Figma and Anthropic as an example. According to media reports, Anthropic surprised its partner by launching Claude Design, which was accused of entering the application layer domain of its ecosystem partners during product expansion, thereby altering the value capture structure. He further pointed out that similar patterns have emerged in the expansion of product lines like Claude Code and Claude Legal, where model capabilities extend upward into vertical application domains. Sacks believes this trend indicates that model vendors are shifting from being "base model providers" to "vertical application competitors," increasing the risk of supplier dependency for enterprise clients. The essence of enterprise AI security, he argued, is not trusting the long-term commitments of model vendors but ensuring optionality and control at the model layer to safeguard their own data and commercial "alpha."