Family offices are investing in AI. Most of them are not actually using it.

Two thirds of family offices now list artificial intelligence as a priority investment theme, according to the latest J.P. Morgan Global Family Office Report. They are putting money into AI-related companies, funds, and infrastructure. But when it comes to using AI inside their own operations, the adoption rate is much lower.

Only 22 per cent use AI for automating manual tasks. Another 22 per cent use it for investment analysis or forecasting. The proportion saying they have deployed AI internally nearly doubled from the previous year, which sounds impressive until you realise it doubled from a low base.

The investment side

Family offices are allocating to AI in three main ways. Thirty-seven per cent are investing in power and resources (the physical infrastructure that AI systems need). Another 37 per cent are in broader infrastructure. And 33 per cent are in AI-enabled healthcare, which has been one of the faster-moving application areas.

The paradox is that many of these offices are investing in AI as a theme while having no real AI capability of their own. They are making bets on the technology transforming industries while running their own operations on spreadsheets and email chains.

The risk side

More than half of family offices in the J.P. Morgan survey said that AI-driven cyber threats are the biggest risk they expect to face over the next five years. Deepfakes, AI-enabled fraud, and automated phishing attacks are the specific concerns.

Only 48 per cent felt confident in their cybersecurity arrangements. That is not a comfortable number when the threat vector is advancing faster than the defences.

Why the gap exists

Three things explain most of it. First, family offices are small organisations. An office with ten or fifteen people does not have a technology team. Implementing AI tools requires someone who understands what the tools can do and how to integrate them into existing workflows. That person often does not exist on staff.

Second, the data is not ready. AI tools work well on structured, clean data. Family office data tends to be scattered across custodians, administrators, fund managers, and internal records. Connecting it all into a format that an AI tool can use is a project in itself, and not a cheap one.

Third, the governance is lagging. Even in offices that want to adopt AI, there is no clear framework for how to use it responsibly. Who signs off on an AI-generated investment recommendation? What happens when the model is wrong? Who is liable? These questions are not hypothetical. They are the reason that many offices are waiting rather than moving.

What will change

The gap will close, but slowly. The offices that move first tend to be the larger ones with dedicated technology staff and a CIO who has the authority to experiment. The smaller offices will follow once the tools become easier to deploy, which is happening but not as fast as the vendors claim.

The real question is not whether family offices will use AI. They will. The question is whether they will use it well, or whether they will bolt it onto broken processes and wonder why it does not help.