Corporate procurement of AI systems is shifting focus from model performance to data retention and security policies. Companies like Nvidia, Palantir, Booz Allen Hamilton, and Novo Nordisk are establishing strict boundaries for third-party AI model operations to ensure compliance with internal data governance.
Providers are adapting to these requirements. OpenAI offers zero data retention (ZDR) options and private safety processing. Google Cloud provides configurations for ZDR deployments on its Gemini Enterprise Agent Platform. Anthropic introduced Enterprise Frontier Safeguards (EFS) in September, allowing customers to store data on their own infrastructure using their own encryption keys.
The financial impact of AI adoption now includes risk management costs beyond license fees. According to a September 2025 PYMNTS Intelligence report, 7% of US enterprise CFOs have deployed agentic AI in live finance workflows, with 5% running pilots.
Organizational readiness remains the primary hurdle for large-scale adoption. An April PYMNTS Intelligence report found that 71% of executives at firms with at least $1 billion in annual revenue cite organizational readiness as the main limitation, while only 11% identify the technology itself as the barrier.
