AI product management · Industry brief

The Corporate AI Governance Report 2025, published by AICDI Global Insights, found that 87% of companies lack a named AI governance framework, only 13% maintain human oversight policies, and just 2.3% have dedicated complaint mechanisms for AI issues. The Partnership on AI issued guidance on May 30, 2025, identifying four critical governance gaps organizations must address: supply chain responsibility controls, end-user terms and conditions, AI assurance ecosystems, and real-time monitoring of autonomous agents. Atos published an AI governance playbook advocating for federated models combining local business-unit autonomy with centralized controls across five domains including governance structures, identity management, data governance, and executive AI use-case review. A Harvard Law Review article analyzed OpenAI and Anthropic's governance structures, questioning whether their capped-profit models and stakeholder boards are sufficient to prevent unsafe incentive structures and ensure accountability. Microsoft, Google DeepMind, and xAI have signed agreements with NIST's Center for AI Standards and Innovation granting U.S. government pre-release access to frontier AI models for national security evaluation, extending a program previously limited to Anthropic and OpenAI.