AI Governance Gaps Leave Enterprise Security Exposed

Enterprises worldwide are integrating AI into their operations at an accelerating pace, from rolling out autonomous agents to embedding AI capabilities into security platforms and internal systems. Yet this rapid deployment outpaces the establishment of protective frameworks and governance structures needed to safeguard infrastructure and customer information, according to cybersecurity professionals. Executives and board-level decision-makers must prioritize cybersecurity and governance considerations before expanding AI initiatives further.

A PwC report from October reveals that organizations lack adequate defenses against attacks targeting AI infrastructure. Among security leaders surveyed, half identified AI protection as their most significant preparedness shortfall—surpassing concerns about cloud security, vendor breaches and ransomware threats. "I actually think we're seeing a growing awareness among business leaders that AI opportunity and AI risk have to be considered together," said Tonya Ugoretz, co-leader of the Cyber and Risk Innovation Institute at PwC. The survey encompassed responses from more than 3,900 business and technology leaders across 71 countries.

Cyber risk leadership

Establishing explicit ownership of AI systems represents a critical governance challenge. Many organizations struggle with ambiguity regarding who oversees AI rollouts and who determines the security measures necessary to protect data and critical infrastructure. "Our findings show there isn't yet one ownership model for AI governance and risk," Ugoretz stated.

The report indicates that fewer than 30% of organizations assign AI accountability to CIOs, CTOs or comparable technology executives. Roughly one-quarter maintain a dedicated AI executive role, while 17% place AI accountability with the CISO. This fragmentation creates uncertainty about who bears ultimate responsibility for managing AI-related cyber threats and ensuring security protocols are embedded before systems go live.

Agentic oversight

Organizations have committed substantial resources to deploying agentic AI tools, frequently targeting productivity gains or product optimization. A pressing concern for cyber risk executives involves the scope of permissions granted to these systems and whether adequate human monitoring exists to track their actual capabilities and actions. "Treat every agent as a new identity with a named human owner," said George Gerchow, chief security officer at Bedrock Data. "An agent inherits the access of whoever it runs as. It never has to break it, it walks through the doors we left open."

According to a third-quarter KPMG report, 49% of organizations have implemented restrictions on autonomous agent access in high-risk scenarios. This represents a significant increase from 28% in the same period the previous year.

Third-party risk

Managing third-party software vulnerabilities remains among the most pressing security concerns facing enterprises. Rather than building proprietary AI solutions, many organizations rely on external vendors for their AI tools. These third-party applications frequently possess extensive access to confidential data and elevated system permissions, enabling direct connections to critical infrastructure. When security teams lack adequate insight into third-party integrations, AI tools may reach sensitive information beyond the scope of existing security systems.

"So, vetting those vendors from both a cybersecurity and a technical standpoint" is essential, according to Pasha Sternberg, a principal and privacy and cybersecurity attorney at Polsinelli.

Shadow AI

Shadow AI—unauthorized or unvetted AI applications operating within organizations—presents an ongoing security challenge. Employees routinely install AI tools to enhance their productivity, develop new competencies or for personal purposes, often without organizational approval. Organizations typically have minimal visibility into how these tools operate or what security risks they might introduce when interfacing with corporate systems or customer information.

The National Cyber Security Centre in the U.K. cautioned organizations last month that shadow AI can create unforeseen vulnerabilities in enterprise environments. Legal professionals have warned that companies frequently underestimate the consequences of unsanctioned tools being deployed by their own workforce. "It's hard to secure and manage what you don't know exists," said Sarah Glover, a shareholder and privacy and cybersecurity attorney at Polsinelli.