The challenges posed to regulators by the rise of AI have been well documented in recent months, but regulators’ own use of AI has attracted far less attention. In this article we take a look at the Advertising Standards Authority’s (ASA) Active Ad Management System and consider what it means to advertisers and media owners, as well as what it might signal about the use of AI by other regulatory bodies in the UK in the future.

The ASA’s Active Ad Management System

Since 2021 the ASA has been working to enhance its AI capabilities in order to increase its capacity to proactively monitor and review ads for compliance with the CAP Code. The result of those efforts is the Active Ad Management System, which replaces the authority’s reliance on reactive manual human reviews with a proactive scanning and monitoring system.

How the Active Ad Management System Works

The Active Ad Management System has greatly increased the ASA’s ability to detect problematic ads. The process works as follows:

• Step One: Ad Capture
  AI technology scans vast amounts of data from both public and private repositories, including social media platforms and search engine results, to capture ads at scale.
• Step Two: Ad Filtering with Machine Learning
  Using advanced machine learning models, the system processes over 3 million ads each month, filtering through them with a combination of image and text recognition. The system’s efficiency continues to improve over time, allowing it to identify the most problematic content.
• Step Three: Manual Expert Review
  The AI system compiles a list of ads that may have breached regulations, which is then presented to the ASA’s compliance team for review. The list is organised allowing the reviewers to see the type of breach and product category concerned, allowing the team to make informed decisions based on the most relevant data.

This AI-powered system has demonstrated remarkable growth: In 2021, the system reviewed 14,000 unique ads monthly; by Q3 2024, that number had increased to 3,486,000, reflecting significant improvements in capacity and accuracy.

Shaping ASA’s Regulatory Approach

The introduction of AI has not only enhanced the ASA’s ability to detect rule violations but has also shifted the way it approaches enforcement. In particular, the authority has adopted a sector-specific and issue-driven strategy to tackle emerging concerns, deciding on a specific theme or issue and then leveraging the AI tool to capture and filter ads that potentially breach the relevant sections of the CAP Code.

Gambling Industry Focus

Since 2022, the ASA has issued multiple rulings concerning the use of sports personalities in gambling advertisements. The regulator expressed concerns that such ads could disproportionately target under-18s in breach of rule 16.3.12 of the CAP Code and rule 17.4.5 of the BCAP Code. By leveraging the Active Ad Management System the ASA was able to swiftly identify which players were being featured in advertising across various forms of media to guide its enforcement efforts.

Similarly, in late 2024, the ASA acted on a series of social slot game advertisements, issuing five decisions on the same day. These games, which simulate gambling without offering real-world financial rewards, were allegedly being marketed in a misleading way in breach of rules 3.1 and 3.3 of the CAP Code. The ASA ruled that several of these ads violated advertising standards due to misleading imagery and terminology suggesting real-money prizes were available.

Aviation Industry Scrutiny

In September 2024, the ASA issued three rulings against aviation companies for misleading "green" claims made in their advertising. All three decisions stemmed from Google Search ads, where airlines had made unsubstantiated environmental claims. This enforcement against a specific form of ad placement suggests that the Active Ad Management System is being utilised to not only look at specific issues under the code, but also specific ad formats.

Key Takeaways for Advertisers

The outcomes of the ASA’s use of AI offers valuable insights into the areas it has prioritised for regulatory scrutiny. For advertisers, this enhanced oversight brings enhanced enforcement risk, and increases the importance of compliance.

• Higher probability of detection: AI has empowered the ASA to identify problematic ads quickly and proactively in a wide range of media, including Google search ads and social media platforms like TikTok. Advertisers should remember to ensure their marketing materials are compliant regardless of where they appear.
• Staying up to date: With increased rulings providing more regular insight into the ASA’s application of the CAP Code, it is important that advertisers keep up to date with where the ASA draws the line for compliance.
• The importance of training: To avoid the disruption and potential negative impacts of an adverse ruling, advertisers should consider regular training for their marketing teams on ASA rules and the potential consequences of non-compliance. 

Broader Trends: The Use of AI by Regulators

The ASA is not alone in its efforts to integrate AI into regulatory practices. The Competition and Markets Authority (CMA) has also signalled its intention to use AI more extensively, having already piloted AI tools to assist in reviewing evidence. The FCA similarly uses AI-powered tools like web scraping and social media monitoring to identify fraudulent schemes, with plans to expand these efforts to improve market surveillance.

What is clear however is that regulators’ use of AI is, at the moment, focussed primarily on building capacity to identify and filter potential wrongdoing amongst massive volumes of data, in an effort to increase the effectiveness and efficiency of the human decision-making stage. 

Utilising AI for regulatory decision-making would come with clear transparency, accountability and fairness concerns but, in the absence of any specific AI legislation, may not be specifically prohibited in the UK. A Private Member’s Bill, the Public Authority Algorithmic and Automated Decision-Making Systems Bill, is currently being debated in Parliament which would restrict public authorities from utilising AI systems where its decisions cannot be effectively assessed and monitored. Even if passed, that would appear unlikely to apply to independent regulators such as the ASA (although the CMA and FCA would likely be closer to being in scope), but it does perhaps set the tone for algorithmic and automated decision-making by bodies with regulatory enforcement powers in the future.