How a government found potential youth and child abuse using AI

The Challenge

The government is responsible for a range of services to support children and youth, one of which is managing reports of abuse or sexual assault against public servants. This can be complex, as information is often reported in different systems, at different times, by different informants. Perpetrators can continue their abuse for years without being stopped if the data isn’t identified, linked, and actioned.

The Solution

We implemented Castlepoint for a government department that had identified systemic failings in its management of reports of inappropriate behaviour by staff in the health and youth services.

Reviews had found that reports made were not acted on for many years, and deficiencies in linking reports together between systems, within and across various departments, allowed abusers to escape notice. Castlepoint was implemented to find key terms related to abuse and misconduct across multiple large business systems and was able to do so with 100% accuracy.

The Result

Castlepoint was implemented on the government network and connected to the first key system within two weeks. The team worked with the Department to develop the taxonomy set, using ontology and regular expression to define all terms in scope. The taxonomy was run across 50,000 database entries over the next month, achieving 99.9% accuracy compared to manual searches.

The taxonomy was then applied to the next system in less than one day, and returned over 60,000 previously unidentified results, identifying multiple potential new cases requiring investigation.

Overall, the project was completed on time and under budget. It resulted in substantial and significant improvements in the ability to detect and stop perpetrators.