Secret agents: Is AI risk being overlooked by both clients and firms?
With all the noise around AI and firms eager to show how they can support clients on their AI journey, it’s hard to believe there could be an area of AI that’s not getting the attention it deserves. But we believe that AI risk is just that.
As clients move agentic AI out of pilots and into live workflows, the opportunity for rogue agents to wreak havoc increases. But leadership attention may understandably be diverted towards adoption as they try to keep up with competitors, while dealing with the latest disruption in their market. Regulation is likely, but it has yet to arrive, and until it has teeth, clients are likely to read this as permission to continue their current approaches to implementation. And this is why we think there is an under-served opportunity in the market.
Clients want speed, but their guardrails aren’t keeping up
There’s a gap between how quickly clients are moving on AI and how ready they are to do it safely.
Upcoming Source research around AI adoption shows that more than a quarter of clients expect to be able to take an AI concept to fully scaled enterprise deployment within six months. The reality may not always match up, but that’s still an admirable ambition, and an indication of just how fast the technology is being incorporated into the daily lives of workforces.
However, our latest Risk & Cybersecurity Market Trends survey flagged a key flaw in this ambition. When asked what kinds of governance they have around AI usage, clients are most likely to say they’ve shared guidance about what tools can be used and what data can be shared. That means they’re essentially trusting staff to understand what’s required and comply. In many cases, they are assuming that the agents being deployed are going to work as planned, but what if one of them turns out to be drifting? Or, in the worst-case scenario, an undercover villain?
We’re not saying that clients shouldn’t trust their workforces, but when we talk to them, it seems there’s a large grey area around best practice. Just one rogue agent is all it takes. Leaders think avoiding this is a technology issue, rather than a risk one, and few seem to be talking about it as a people issue.
Firms are selling speed over safety
Firms aren’t necessarily helping with this situation either. When clients look for AI adoption support, they think about process optimisation and workforce reskilling rather than risk & ethics, and firms don’t seem to be dissuading them from this, instead talking up the speed at which they can deliver solutions and start to derive benefits for clients.
Sell AI risk as an accelerator, not a brake
So how can firms tap into this potentially lucrative market? By building on the signals that are coming from clients and reframing AI risk as something that clients need to buy to move faster.
Clients know they should be looking at this. Help with implementing AI is a common reason for clients using more risk consulting, and technology risk is an area they know they are going to need more support. AI risk might not be given a place at the top table, but it probably doesn’t need to fight for a place on the leadership agenda. Instead, it needs to be recognised by those outside the technology and risk departments as an essential part of every AI project.
What can firms do next?
We track, forecast, and size changing demand for AI risk services across the market. Get in touch to find out where AI risk demand is strongest among your clients, and how big the opportunity could be for your firm.