New Delhi, Nov 3 (NationPress) Indian corporations are swiftly advancing in the realm of AI, driven by global trends, competitive pressures, and breakthroughs in GenAI technologies, as revealed in a recent report. AI applications now span customer engagement, operational optimization, and essential processes across various sectors.

However, the adoption of AI remains uneven, with only 15 percent of organizations achieving comprehensive enterprise-wide deployment.

"Despite the ongoing expansion of AI, governance is not evolving at the same speed. In numerous organizations, the AI infrastructure is growing quicker than the necessary governance, security, and ethical safeguards, resulting in significant gaps in accountability and risk management," stated Alvarez & Marsal (A&M), a global professional services firm, in its findings.

Furthermore, governance maturity is still limited, even as usage increases.

While 60 percent of organizations have established basic governance or acceptable-use policies, only 19 percent have conducted thorough risk assessments, and 81 percent lack full visibility into the monitoring and governance of their AI systems.

With numerous AI initiatives developing in isolation, accountability and standards vary significantly, particularly when third-party and in-house models coexist.

The report underscores the necessity for integrated, organization-wide governance frameworks that promote transparency, oversight, and clear role ownership.

“AI is now more deeply integrated into business processes and decision-making systems than ever before. India’s AI potential is considerable, but its long-term benefits hinge on how effectively organizations govern and secure the systems they implement," remarked Dhruv Phophalia, MD and India Lead - Disputes and Investigations, Alvarez & Marsal.

Those who invest early in these foundational aspects will be optimally positioned to realize the full economic and competitive advantages of AI, he added.

According to the report, while the principles of responsible AI are widely recognized, their practical application remains limited.

Less than 20 percent of organizations have implemented mechanisms for explainability, bias detection, or fairness, and 60 percent do not have any formal processes to validate model integrity.

Data governance presents similar deficiencies, with only 26 percent having integrated data masking and PII scanning in their AI workflows, and 60 percent performing no structured validation of datasets.

The report also pointed out that as increasingly complex AI models are deployed, security throughout the AI lifecycle will become critical.

While 52 percent of enterprises maintain secure development environments with basic controls, less than 30 percent engage in penetration testing or red teaming, and only 19 percent have measures in place to detect data poisoning during model training.