Last week was an eventful one for the wealth management industry.

First, Anthropic launched Claude for Financial Advisors. It arrives with ready made workflows for meeting preparation, portfolio review, estate and tax briefings, post meeting notes and client onboarding. A few days earlier, OpenAI had released its own version for the industry.

Then came the FutureProof festival in California. Five thousand people, from wealth managers and private banks to multi family offices and the technology firms that serve them. The main topic was how to accelerate the use of AI across the industry.

Wealth management has lagged on technology for years. Some of that is regulation. Most of it is that the work is genuinely hard to scale with software. Understanding one family's objectives, tax position, liquidity needs and plans takes time. It requires human judgement and a relationship, and that time does not compress.

Assembling context from fragmented data sources is exactly what AI is good at. There is no question it buys advisers time back, so they can spend more of it with clients, where the human part actually matters.

But the speed of this transformation has outrun the regulation, the compliance and the governance around it.

The client was not in the room

Five thousand professionals, and a great deal of discussion about connecting client data to AI systems. As far as I can tell, no part of that conversation involved the clients whose data it is.

Firms are signing up. Data is being connected. I have not seen much evidence of firms going back to clients to explain what is changing, or offering them a choice about it.

I do not have hard data on how widespread that is. But I have not yet seen a firm say publicly: here is what we are adopting, here is what happens to your information, and here is your option to decline if you are not comfortable with any of it.

Consent is being assumed under existing data policies.

Why this one feels different

Think about your relationship with your doctor.

Your doctor has the same problem. Fragmented records, time lost assembling context, less time with you. An AI layer across your medical history would genuinely help. Previous conditions, medications, what was discussed last time, all of it in one place before you walk in.

Now imagine your doctor mentioned at the end of an appointment that they had connected your health record to a new system so they could serve you better.

Most people would have a strong reaction to that. Not because an AI is making the diagnosis. It doesn't; the physician remains responsible for interpreting the output. The objection is mostly about the data.

Where does my data sit? Who can see it? Is any of it used to train anything? How good is the security? What could go wrong? And the question that really matters: what happens if my health data reaches an insurer, an employer or a government?

Those are legitimate concerns, and we do not yet fully understand the consequences. Regulators have reached the same conclusion. Under the EU AI Act, uses of this kind sit in the high-risk category, or are prohibited outright.

The consequences are not theoretical either. They show up as the cover you are offered, the price you are quoted, and the things you are quietly excluded from. That is the doorway to social scoring, to discrimination, and so much more.

What surprises me is how rarely financial data gets treated the same way.

A full picture of a family's assets, entities, income, tax position, liquidity, beneficiaries and estate plans is every bit as personal as a medical history. In some places it is considerably more dangerous to have exposed, because the consequences run well beyond a higher insurance premium.

The timing

In the same week, Dario Amodei, Anthropic's chief executive, published an essay, "We Must Pace the Frontier", calling on the industry to slow down. Sam Altman agreed publicly within days. Both were responding to incidents in which AI agents acted autonomously and broke into systems their own developers had not intended them to touch.

So a single week contained a serious warning from the people building this technology, and a large push to connect that technology to client financial records across an entire industry.

When the people with the most information are publicly uncertain about the pace, it is worth thinking about it while your own data is being plugged in.

My take

The AI industry is not going to slow down, and it is not going to ask your permission. That is simply how technology adoption works, and the commercial pressure behind it is enormous. The genie is out of the bottle, and this technology is going to shape everything.

I have stopped trying to work out whether we can trust it. I have started planning how we benefit from it while keeping sovereignty and control over our own data, which is among the most valuable assets any of us hold.

So here is my challenge for this week. Look at where your data currently sits, and audit the whole architecture. Who are your vendors and advisers, and what technologies are they using to serve you. Start by asking each of them for their AI policy, so you can understand your rights, your options, and where your information might travel.

Then ask three simple questions.

Which AI systems have access to my information?

Where does my data sit, and is any of it used to train AI?

If I moved firms tomorrow, what would come with me?

A good firm will have thought about all three already, and will have a written policy it can share with you.

If it does not, that is your answer.

Until next week.

— Amin Naj