Searchlight Cyber adds AI capabilities to summarize dark web posts and threads

Searchlight Cyber has launched new AI capabilities within its dark web investigations platform, Cerberus.

The new Cerberus AI Insights feature summarizes dark web posts and threads to help analysts to quickly triage information and focus on the dark web content that is most relevant to their investigation and operational scope.

Searchlight Cyber’s dark web investigation platform Cerberus contains more than 15 years of historic dark web data and continuously captures new posts from dark web sites, marketplaces, and forums. This information is automatically categorized, indexed, and made searchable – making it an invaluable investigative tool for law enforcement and cybersecurity analysts.

However, even with this legwork undertaken by Searchlight, until now it was still a manual process for investigators to read through dozens or even hundreds of messages and posts to identify whether a dark web thread contained any valuable intelligence.

Searchlight’s latest AI capabilities release is designed to further increase the speed of investigations by helping analysts to quickly determine the relevance of conversations and threads to their objectives. This is achieved by providing AI overviews that summarize the contents, key statistics, and sentiment of dark web conversations. Key features of Cerberus AI Insights include:

• Thread summaries – AI summarization of long forum threads into an easily digestible overview that helps analysts to quickly determine if the conversation requires further investigation.
• Threat categorization – Identification of threats within a forum thread by topics – such as drugs, arms, or cybercrime – to determine whether it relates to their area of investigation.
• Forum thread insights – Statistics on each thread including the number of active users, the total message count, and lifespan of the conversation, helping investigators to quickly determine factors like the popularity of a post.
• Sentiment analysis – To detect shifts among dark web users from escalating violent rhetoric in a forum, to gauging the cybercriminal community’s reaction to the launch of a new ransomware strain.

“One of the challenges for dark web investigators is the vast amount of information they need to sift through in order to find the golden nuggets of intelligence that make the difference. Dark web forums posts are often long and complex, with a number of actors discussing various topics over days or weeks. It requires a lot of resources for human analysts to keep up with these posts – with new ones being added all the time – but is another great use case for AI, which has proven itself powerful in extracting insights from large amounts of unstructured, textual data,” said Dave Osler, Head of Product at Searchlight Cyber.

This latest update follows the previous deployment of bespoke AI-powered language translation for the top 10 languages on the dark web. This feature enables investigators to translate content within the platform at a much higher level of reliability, as the model is trained on Searchlight’s dark web data and therefore mitigates factors like misinterpreting hacking slang.

“For every upgrade we make to Cerberus, we are asking ourselves how we can give law enforcement and cybersecurity professionals the upper hand against criminals operating on the dark web. These new AI capabilities – like our existing AI translation feature – are all about helping our customers to more quickly extract the intelligence from the noise. The Cerberus AI Insights is in saving our customers valuable time in identifying relevant threats and high-value intelligence,” concluded Osler.