A wellness label cannot settle what a health app promises
CAP guidance and the Do Health ruling show why health-app marketers must distinguish product classification from evidence for the benefits advertised.
Read the noteNotes
Observations on expert work, intelligent systems and regulated marketing
-
If AI does the first draft, where do future experts learn?
Anthropic’s workforce research raises a problem for regulated teams: automating routine production can also remove opportunities to develop professional judgement.
-
Human oversight has to survive the AI being wrong
A neonatal diagnosis study highlights differences in how reviewers handle AI errors, suggesting a more demanding test for human oversight.
-
What 72,400 ads reveal about compliance monitoring
The ASA’s enforcement reports expose the work between detecting suspect advertising and establishing that an advertiser has corrected it.
-
When AI becomes the advertising claim
ASA rulings on Finelo and AI Scorecast show why AI advertising must substantiate promises about ease, reliability and financial outcomes.
-
Enterprise AI needs to know which version of the organisation to believe
OpenAI’s Atlassian partnership and Infor’s research raise a question about enterprise AI context: which business information can guide action?
-
The claim can stay the same while the truth beneath it changes
The FTC’s Greenway complaint highlights a pricing gap, with lessons for keeping marketing claims connected to current product facts.
-
Compliance can see the problem and still lack the power to change it
The FCA’s corporate finance survey highlights why compliance needs authority and early involvement to influence commercial decisions.
-
AI governance in financial services needs evidence from the work
The FSB’s 12 sound practices challenge compliance teams to establish whether an AI application performs reliably in context, then keep testing that judgement.
-
What has been approved when every customer sees a different message?
Zango’s financial services essay collection raises a question for AI personalisation: how does approval hold when each customer sees different copy?
-
Finding the information is becoming the easier part
S-RM’s survey suggests AI speeds up compliance research while context, credibility and relevance still require expert judgement.
-
Fourteen compliance systems still need an answer about authority
MCO’s survey reports an average of 13.8 compliance systems, raising a question about which records should govern a financial services decision.
-
A convincing compliance answer is not a complete assessment
A study of 12 EU AI Act compliance checkers finds gaps between risk classification and guidance an organisation can act on.
-
AI approval needs to survive the next change
The UK’s healthcare AI reforms point towards lifecycle assurance, with lessons for keeping approvals valid as systems and evidence change.
-
The advert does not end where the landing page begins
The ASA’s Juniper ruling shows why intervening pages do not necessarily stop a customer journey promoting prescription-only medicines.
-
An AI agent can reach the data. That does not settle its use.
The ICO’s agentic AI consultation highlights why access to personal data does not settle its use, retention or reuse in marketing.
-
brett and Modern Lender launch State of AI in Marketing
The first edition of brett’s new research programme will benchmark how AI is changing marketing work across UK mortgage lending.
-
When the prompt omits the rule
The Midnite ruling shows where generative marketing needs controls: creation, review and publication.
-
Why questions don't always create curiosity
Across four studies, question-framed titles drew less engagement because readers expected them to be less informative.
-
When criticism helps the review but hurts the reviewer
A negative review can feel more useful while its author seems less credible, depending on what the audience expected.
-
The option you do not want can still change the decision
A less-preferred option can reduce the appeal of a wider proposition, even when customers are free to ignore it.
-
Why voices that sound like ours may earn more trust
New research suggests similarity in vocal timbre can shape competence-based trust and persuasion.
-
Advertising on ChatGPT: what should marketers approve?
An advert can pass content review and still be unsuitable for the conversation in which it appears.
-
Where marketing’s AI productivity gains go
McKinsey respondents widely report individual AI productivity gains, while far fewer report organisation-level financial impact.
-
The age of AEO belongs to interesting brands
AI is changing how content is created and discovered but what comes out still depends on what a brand puts in.
-
What should an AI agent remember from expert work?
Useful organisational intelligence depends less on retrieving previous decisions than on preserving the expert criteria, reasoning and contextual limits that made them valid.
-
How do you approve an ad when the context is private?
Regulated advertisers need to show that contextual ad controls worked without seeing the private conversation behind each placement.
-
Nine marketing compliance platforms for financial services (2026)
Compare nine platforms across rule intelligence, pre-review, approvals, asset management and live monitoring.
-
The Consumer Duty boundary is becoming a marketing question
As the FCA redraws the boundary around retail protections, marketing teams will need to pay closer attention to context.
-
From AI risk frameworks to accountable decisions
New research reveals where AI risk classification and tooling fall short: accountability, human oversight and wider governance.
-
Employee voices need systems, not scripts
Employee-generated content can give regulated brands credible expertise and greater reach, but scaling it depends on guardrails that help people exercise judgement—not copy that removes it.
-
FINRA proposes moving beyond the approval queue
FINRA wants to replace the default principal pre-use approval requirement with risk-based supervision. The opportunity is faster communication; the harder task is proving why each asset received the scrutiny it did.
-
Article 50 turns AI transparency into a workflow decision
The EU AI Act’s transparency rules have applied since 2 August 2026, making the relevant obligation, accountable owner and supporting evidence part of the review.
-
The Exhale Coffee ruling: evidence is not permission
Exhale Coffee supported its marketing with research, laboratory testing and medical expertise. The ASA ruling shows why none of those things, alone or together, creates an authorised health claim.
-
The EU AI Act is moving into company workflows
Research on nearly 3,000 companies shows how the EU AI Act is shaping governance beyond Europe and the evidence behind responsible-AI claims.
-
Consumer Duty monitoring has to show what changed
The FCA’s latest review finds that collecting management information is not enough. Firms need a traceable path from customer evidence to decisions, action and a measured result.
-
The PM’s AI Taskforce faces the second-use-case test
Lord Vallance has been given direct authority to drive the government’s AI strategy. The harder test is whether evidence and judgement from one deployment make the next easier to govern.
-
How US bank marketers are putting AI to work
Data reveals where bank marketers are applying AI, how they access it and why the most valuable applications remain the hardest to implement.
-
Marketing is becoming the test of organisational AI
Research on AI marketing adoption shows why regulated teams need to examine the workflows around individual tool use.
-
AI adoption is rising faster than organisational capability
ONS data shows AI use spreading quickly across UK business. The more valuable task for marketers is turning isolated use into well-governed organisational capability.
-
The missing layer in agentic marketing
BCG's 2026 CMO Survey shows that adopting AI tools is not the same as transforming marketing. In regulated industries, the missing layer is organisational intelligence.
-
The creator economy is coming for regulated professions
Expertise is increasingly discovered through content. Regulated professionals have a significant opportunity, but publishing at scale changes the demands on judgement, evidence and review.
-
AI governance in financial services needs a human test
The Mills Review exposes a weakness in many AI strategies: putting a person in the loop is not enough. Accountable delegation requires a far more deliberate operating model.
-
FCA regulatory strategy: the end of compliance at any speed
The FCA wants firms to make faster, better-evidenced decisions about risk. For marketing and compliance leaders, that demands a better operating model, not lower standards.
-
AI in financial services has moved beyond the pilot problem
The UK’s financial services AI plan reveals that the hardest part of adoption is no longer proving the technology. It is turning regulatory principles into repeatable decisions.