85% Indian finance leaders face AI ROI pressure, 27% lack accountability

For AI practitioners, the survey reveals a stark tension between the push to deploy agentic systems and the neglected need for robust governance. In financial contexts, where algorithmic decisions directly affect compliance and revenue, the lack of explainability and unclear accountability chains are recipes for disaster. This data underscores an urgent call for the AI community to design agent frameworks with built-in audit trails, human-in-the-loop reviews, and transparent decision pathways—before regulatory backlash or high-profile failures force a painful reckoning.

A recent survey from tax compliance software firm Avalara exposes a glaring contradiction in India's corporate finance landscape: while artificial intelligence agents are being deployed at breakneck speed, the governance frameworks needed to manage their risks are being systematically neglected. The survey, which polled CFOs and senior finance leaders across India, the United States, the United Kingdom, and Australia, found that 85% of Indian respondents face moderate or significant pressure to demonstrate returns on their AI investments. Yet only 8% said their organizations prioritize governance over speed, and 27% reported that accountability for significant AI errors is either unclear or assigned to no one at all—the highest proportion among the four markets surveyed.

“

Meanwhile, 24% said their organizations had not updated internal controls in the past year to account for AI agents that are now taking or recommending actions, and only 28% reported those controls had been reviewed by IT or cybersecurity teams.

This rush to automate without adequate safeguards is not just a theoretical concern; it carries concrete risks. Ten percent of Indian finance leaders admitted they could not confidently explain an AI agent's actions to a regulator or auditor. Meanwhile, 24% said their organizations had not updated internal controls in the past year to account for AI agents that are now taking or recommending actions, and only 28% reported those controls had been reviewed by IT or cybersecurity teams. The numbers paint a picture of a high-stakes experiment where financial processes are being handed to algorithms that even their overseers cannot fully explain or responsibly govern.

The implications extend far beyond mere compliance box-ticking. AI agents in finance influence tax filings, invoicing, fraud detection, and transaction approvals—areas where errors can quickly escalate into regulatory penalties, financial misstatements, and reputational damage. Dulles Krishnan, Vice President and General Manager of India Operations at Avalara, underscored the danger: 'Running new AI tools on outdated compliance policies is a massive blind spot. CFOs in this market need to ensure their risk frameworks are actually updated to monitor automated decisions before an auditor comes knocking.' His warning highlights a systemic oversight: many organizations treat AI as a plug-and-play productivity tool rather than a transformation that requires parallel evolution in oversight, audit, and accountability structures.

India’s fervor for AI-driven efficiency is understandable. The country’s finance sector is among the world’s fastest digitizing, and pressure to cut costs and improve accuracy in a high-volume, complex tax environment is immense. But the survey data suggests many leaders are caught in a classic innovation trap: they chase the upside of technology without fully accounting for the downside. The fact that only 34% of AI controls have been reviewed by risk or compliance teams—and even fewer by IT or cybersecurity—means a significant portion of agentic decisions may be running unchecked, with no clear audit trail and no designated human owner when things go wrong.