Credit: Anthropic Anthropic has launched Claude for Financial Advisers, wiring the chatbot into portfolio and research tools from BlackRock, Vanguard, Charles Schwab and iCapital, and says it will not give investment advice. Under MiFID II advice turns on a five-part test in which suitability can be implied rather than stated, and ESMA has told firms that using AI changes none of their obligations. Anthropic is pitching financial advisers a version of Claude wired into analytics and risk tools from BlackRock, Vanguard, Charles Schwab and iCapital, Bloomberg reported. Claude for Financial Advisors promises to speed up research, administration and portfolio oversight.
Executives at both companies called it one of Anthropic’s biggest moves into finance so far. It follows the financial services agents the company shipped earlier, and lands a week after OpenAI launched its own product for bankers and equity researchers. The pitch is capacity. Jonathan Pelosi, Anthropic’s head of financial services, said the adviser community is shrinking and that “these people are retiring, there’s not a ton of them“.
BlackRock says it manages about $300B in the model portfolios advisers increasingly buy off the shelf. Advisers want to outsource, its wealth advisory head Jaime Magyera said, and the opportunity is capacity rather than information. Then Pelosi drew the line. “You won’t explicitly get investment advice from Claude. We reserve that judgment for the experts,” he said.
In Europe that sentence is a legal boundary rather than a disclaimer. Investment advice under MiFID II means a personal recommendation, and the European regulator’s test for one has five cumulative parts. One asks whether an instrument is presented as suitable for the person. ESMA says that can be done in an explicit or an implicit form, which is precisely the word Pelosi’s sentence rests on.
The same briefing treats references to an adviser as covering automated and semi-automated systems, and ESMA told firms in May 2024 that using AI changes none of their obligations to act in a client’s best interest. The AI Act is not the instrument here. Its high-risk list reaches credit scoring and the pricing of life and health insurance, and says nothing about investment advice or portfolio management. A European firm has already tested the edge of this.
Scalable Capital connected EUR 60B of client assets to ChatGPT and Claude in August, letting customers analyse portfolios and trade through them. That is a regulated broker exposing its own clients to consumer assistants. Anthropic is selling a professional tool to the adviser instead. It therefore sits one step back from the client.
The question European supervisors will ask is what the adviser does with the output.
Leave a Reply