The UK Financial Conduct Authority (FCA) is preparing to reassess the complex-product risk tests used by brokers to screen retail clients. As generative AI can now provide instant answers to questionnaire questions, regulators and market participants are questioning whether passing a test really shows that a client understands the risks of contracts for difference and other leveraged products.
The issue goes directly to how UK retail trading platforms assess clients, document the sales process and meet their investor-protection responsibilities. If existing questionnaires show only that a client has found the correct answers, rather than understood margin, leverage and potential losses, the tests may no longer perform their intended screening role.
Which products are covered by FCA appropriateness tests
In the UK, FCA-regulated brokers generally need to carry out an appropriateness assessment before offering complex or high-risk products to clients. Depending on the applicable rules and product structure, these may include contracts for difference (CFDs), complex leveraged exchange-traded products, restricted mass-market investments and eligible cryptoassets.
An appropriateness assessment is not the same as a suitability assessment. The former primarily examines whether a client understands the product they intend to buy. It does not determine whether the investment fits the client’s financial position, investment objectives or overall portfolio. Passing the test therefore does not mean the product is appropriate for that client’s risk tolerance.
In some cases, a broker may still issue a risk warning and execute an order after a client who failed the test explicitly asks to proceed. For restricted mass-market investments, however, firms must confirm that the product is appropriate before accepting an order. Multiple failed attempts may also trigger a waiting period.
Generative AI challenges online questionnaires
Traditional online tests have often allowed clients to search the internet for answers after seeing the questions. Generative AI lowers that barrier further: a client can send the question directly to a chatbot and enter the response into the questionnaire.
Brokers and other market participants involved in FCA discussions have noted that a passing score may reflect a client’s ability to find information rather than their actual understanding of investment risks. The gap is particularly important for leveraged products.
For example, a client may be able to define “margin” correctly without understanding how leverage can magnify losses when the market moves against the position. Answering a single terminology question also does not establish that the client understands position volatility, additional funding requirements or the practical consequences of expanding losses.
Brokers apply different testing standards
Firms currently have considerable discretion over both the content and administration of their tests. Some brokers use multiple questions to examine a client’s understanding of how a product works, while others rely on questionnaires that are closer to a formal tick-box exercise. Whether a client passes may therefore depend on the platform used, rather than solely on their grasp of the product.
Firms also differ in deciding which instruments qualify as “non-complex” products and can be exempt from testing. Even when clients buy similar products, they may face different levels of scrutiny at different brokers. That variation makes it harder for the FCA to assess the consistency and practical effectiveness of the tests.
FCA plans further discussions from 2027
The FCA continues to regard appropriateness assessments as an important investor-protection measure. Only a small number of participants in the feedback supported removing or completely replacing the current framework. The regulator is instead leaning towards examining how the existing model operates, what harm could result from weaknesses in the system and whether other screening options are workable.
The FCA plans to begin further discussions with brokers, consumer groups and other relevant parties in 2027. The regulator has not yet proposed a specific replacement model or published a detailed timetable for new rules.
Until the review is complete, brokers remain responsible for how their questionnaires are designed and used. The focus will not be limited to whether clients complete a test, but also whether it can identify their understanding of complex product mechanics and potential losses. For retail clients, passing a questionnaire with help from AI is not the same as understanding the risks involved in leveraged trading.