Build or Buy: The Different Approaches to Adverse Media Screening 

Identifying risk in your client portfolio is a huge ongoing challenge, so one of the most critical questions Chief Compliance Officers (CCOs) must answer is whether to build or buy technological tools for negative news screening.

While building a tool from scratch gives an organisation total control over its development and use, it can become a burden on your time and resources. It demands specialist technical support, significant time for testing and deployment, and constant monitoring to keep it up to date with the latest technological advancements in AI and machine learning.

On the other hand, adopting a tool from a specialist vendor may reduce your level of control, but it offers technical expertise, ongoing support, and advanced, AI-led software that minimises disruption and boosts long-term efficiency.

Build: What are the limits of an in-house solution?

Many organisations will opt for an in-house solution. This provides the maximum level of control over the software, allowing firms to develop it according to exacting specifications.

But they also come with major limitations:

• They can quickly become outdated with rapid technological advancements, especially surrounding artificial intelligence (AI) and machine learning, necessitating expensive and time-consuming ongoing development.
• Solutions that fall behind in technology can produce overwhelming false positives, expensive backlogs, and poor client outcomes. 
• A lack of specialist in-house resources means tools may not perform optimally, such as with over-reliance on ‘fuzzy matching’, which often produces poor outcomes.
• Sifting through false positives to account for non-optimal performance can be a lengthy task and a waste of precious human resources.
• Organisations may lack the resources to promptly service an in-house system in case of an outage or other technical difficulty.

Buy: What are the advantages of an AI-led solution?

Sourcing an AI-led solution from a specialist vendor is the obvious alternative. Although many organisations may be wary of outsourcing a critical operation to a third party, they offer the kind of expertise and ongoing support some firms lack on their own.

An AI-led solution can:

• Dramatically reduce false positives, provide intuitive and easy to understand analysis, and support a more robust screening capability.
• Harness enormous amounts of unstructured data to produce new insights on customer risk, dramatically improve results, cut down on false positives, and overcome inefficiencies.
• Give compliance teams complete oversight and control during investigations, cutting out the unnecessary middleman and ensuring more accountability, transparency, and efficiency.
• Scale without necessarily needing to hire more staff members. This scalability means that AI-led tools have a positive return on investment, making the compliance function a growth enabler rather than a cost burden.
• Remove redundancies and improve overall efficiency, allowing teams to focus on real instances of financial crime risk.

Furthermore, vendors specialising in AI-led technology for anti-financial crime have the resources and expertise to concentrate more effectively on technology advancements, specific regulatory requirements, and producing tools with a good and proven user interface. They also have clear roadmaps for improving their systems, often based on user feedback.

Screening Innovation: Labyrinth Screening, featuring AI Risk Profiles

CCOs must tackle a whole range of challenges when it comes to screening. Customer data can be limited and problematic, while media data can be noisy and imprecise. Many screening methods generate a large number of false positives, struggle to achieve accuracy at scale, and put a significant time burden on analysts.

Ripjar’s screening solution features AI Risk Profiles which are designed to address these challenges directly, saving analysts time and increasing accuracy in customer screening.

Data from both structured and unstructured sources is reviewed to build individual profiles for people and organisations, with advanced natural language processing extracting the most relevant items necessary to give a clear and complete view of relevant risks as quickly as possible.

AI-powered multilingual name matching and entity resolution are used to overcome screening challenges such as common or high profile names, helping ensure your organisation’s regulatory compliance by identifying risks other screening methods might miss.

This approach also captures a huge number of secondary identifiers – such as dates of birth, nationalities, locations and roles – from unstructured text.

This vast expansion of context leads to richer data and better recall. Standard watchlists are also enriched with these additional properties, improving sanctions and PEP screening accuracy.

80% of AI Risk Profiles contain secondary identifiers, which is key to reducing false positives. Testing has shown that there can be as much as a 91% reduction in false positives alongside a 5% improvement in recall.

By aggregating these properties across millions of articles, Ripjar’s screening tool can assign identifiers to entities at a scale which is simply not possible in human-curated profiles, and at an accuracy not achievable with article-based risk evaluation.

---