.png){kind=logo}
AI is rapidly moving from experimentation into business operations. Organisations are using AI for research, content, decision-making and several internal processes. But as adoption grows, companies also need to address privacy, compliance, accountability and trust.
In this episode of the Analytics Insight Podcast, Priya Dialani speaks with Vikram Jeet Singh, Partner at BTG Advaya, about the growing importance of AI governance and data privacy. Singh, who specialises in IT, data privacy, online content regulation, e-commerce and AI regulation, explains why businesses need clear guardrails and human oversight as they deploy AI. Here are the key excerpts:
I think innovation and regulation have to go hand in hand. AI is not just about the output you get from a model. The bigger question is whether you trust that output and how you actually use it.
For businesses, that means deciding where AI can be used and where it should not be used. You need clear guardrails around acceptable and unacceptable use. Organisations are still figuring this out, but I believe these guardrails should be established sooner rather than later. At BTG Advaya, we aim to use AI for things such as legal research, finding case law and precedents, preparing presentations and creating certain marketing materials. But we have clear boundaries.
I would start with a simple AI policy or governance document. There is no prescribed length or format. What matters is that an organisation thinks through where AI fits into its business and where it does not. You need to define how employees can use AI, what information they can put into these systems, what decisions can be supported by AI and where human intervention is required. Governance should also be built into the AI development lifecycle from the beginning rather than being added after deployment.
They are closely connected, but I would not treat them as the same thing. Solving for AI is not solving for data privacy, and the other way around. We already had privacy risks before AI became widespread. These include unauthorised use of information, covert data collection, data brokers and the use of biometric or video data. AI can amplify some of these risks. Deepfakes and synthetic content create another layer of concern. So, when companies build their privacy framework, they should also understand how AI is being used across the organisation.
The first thing I would do is create an inventory. You need to know what AI models and tools are being used across your organisation and what they are being used for. The second step is to create the policy and establish the guardrails. Third, you need human responsibility. I don't think everything can or should be automated. There has to be a human responsible for important decisions. AI governance is not a one-time exercise. AI changes very quickly, so organisations need regular reviews and, where appropriate, board-level reporting.
AI governance should come from the top because AI can affect the entire organisation. Boards and CEOs need to understand what AI is being used for and what risks come with those deployments.
For me, four things are particularly important: human oversight, continuous testing, transparency and risk awareness. There should always be someone responsible for the final output. AI systems also need to be monitored because their behaviour and capabilities can change. Organisations should be transparent with stakeholders about how AI is being deployed and how data is being handled. The objective is to understand the risks, put the right controls in place and use the technology responsibly.
Listen to the full discussion on the Analytics Insight Podcast.
