Enterprises Prioritize AI Accountability Over New Tools

According to a Jitterbit report (press release via GlobeNewswire, republished by The Manila Times), enterprises now place AI accountability at the top of evaluation criteria for new AI tools, with 47% of respondents naming it the single most important factor. The report states companies currently run an average of 28 AI agents and expect that to rise to 40 within the next year, a 43% jump. The report also says 78% of AI projects are delivering real business value, yet 95% of enterprises are holding back on scaling because of security concerns. The report projects a 48% increase in agent deployment among larger organizations by 2027 and highlights risks described as agent sprawl, "God-mode" access to sensitive databases, and exploitable gaps from AI-generated code.

The press release highlights operational risk vectors rather than model architecture changes. It notes uncontrolled agents accumulating high-privilege access and that existing tooling misses some exploitable gaps about one-third of the time, per the report. The release includes direct comment from Jitterbit President & CEO Bill Conner: "Organizations have seen the value; now they just need to scale with AI accountability and security in mind -especially with agent sprawl and agent contamination as real threats to today's enterprise." The press release frames the findings around automation, integration, and accountable AI practices.

Industry observers: Organizations deploying many agentic systems commonly face governance and lifecycle problems as agent counts grow, including inventory, credential sprawl, and drift between intended and actual agent behaviour. For practitioners: standardized audit trails, least-privilege access controls, and deployment-time guardrails are frequently recommended to address similar risks in other enterprise deployments.

Editorial analysis: The report's combination of high reported value (78%) and high reluctance to scale (95%) highlights a common adoption pattern where security and governance lag behind pilot success. This pattern raises operational questions for security teams, platform engineers, and procurement, especially as agent counts accelerate and regulatory attention to data sovereignty and access controls increases.

• •Whether enterprises adopt centralized agent registries, stronger identity/privilege controls, and runtime monitoring.
• •Vendor responses: new features for auditability, policy-as-code, or data-locality options.
• •Regulatory or litigation developments tied to data access and AI-driven actions that could change tooling requirements.