InMarketing Today

The Financial Conduct Authority has called on artificial intelligence providers to explain the limits of the financial information and advice their models generate, City AM reports today. Jessica Rusu of the regulator said in a speech today that the technology 'can significantly widen access to financial tools and information' but 'can also blur the boundary between regulated and unregulated activity'. The report says 40 per cent of young investors mistakenly believe chatbots are regulated by the authority, leaving them unable to get help or compensation when advice goes wrong. Rusu also criticised regulated firms for building tools that lack a defined need, exclude vulnerable consumers or have overstated capabilities.

Regulated status is the one thing a generic chatbot cannot offer, and the report suggests many customers cannot see the difference. If four in ten young investors cannot tell, authorisation is a brand asset worth stating. A disclaimer from a chatbot maker will help at the margin; a regulated firm that makes its own status visible does the work in advance. The criticism of overstated capability is a marketing point too: any claim you make about your own artificial intelligence will be read by a supervisor as well as a customer.

Say it where the advice is given: inside any artificial intelligence feature you offer, state that the service is regulated, by whom, and where a customer can complain. Then check every public claim about your own artificial intelligence tools against what they actually do, and cut those you cannot show. If you sell to regulated firms, expect governance, human oversight and monitoring to come up in the sales conversation, since the regulator named all three.

JPMorgan announced today that it has brought JPM Coin, its dollar-denominated deposit token, to Ethereum for institutional clients, The Block reports. The token, first launched on the Base network in November 2025, represents the bank's existing deposits on a public blockchain and now supports peer-to-peer transfers across the Ethereum Virtual Machine ecosystem. JPMorgan says its Kinexys platform has handled more than $4tn in cumulative transactions, with average daily volume of $7bn.

On a shared network, the issuer's name carries the weight. JPM Coin runs where other tokens run, so what separates it is the claim behind it: a representation of existing deposits at a named bank. That is a brand position as much as a technical one. Treat the figures with care: they come from the bank's own statement, and the $4tn covers the whole Kinexys platform, not this token.

If you issue or distribute anything tokenised, answer in one plain sentence who stands behind it and what the holder is owed, and put that sentence where the product is described. If you sell to banks, find out whether your product works with the Ethereum Virtual Machine, which Matthew Dawson of Ethereum Institutional says the Kinexys platform is built on.

Social Media Today reported yesterday that Meta has set out how businesses are using Muse, its artificial intelligence agents, based on examples shared in a press release late last week. The most common request is to assess which ads work and why. Others include building a campaign structure from a creative brief, chaining Muse's Meta Ads connector with Shopify, Canva and HighLevel into one automated funnel, and setting up product catalogues and pixel tracking. Meta launched Muse for small and medium-sized businesses late last month.

The most common request is a measurement question, which puts the platform's own agent in the position of grading the platform's own advertising. Meta gives no usage figures, results or customer names, and the examples come from its press release, so this is positioning, not evidence of performance. For a financial services marketer the signal is still worth noting: small businesses are being taught to ask an agent for a first view on campaigns and cashflow, one they might otherwise seek from an agency, a bookkeeper or an adviser.

Check results from any Meta campaign against your own data, not only the agent's reading of them, and keep the pixel and catalogue set-up in accounts you own. If you sell to small businesses, write your product descriptions, pricing and eligibility in plain, consistent terms, so that an agent summarising you for a customer has accurate text to work from.