Google Threat Intelligence to use Gemini AI to bolster cybersecurity

Google recently unveiled a new cybersecurity product named Google Threat Intelligence that combines the expertise of its Mandiant cybersecurity unit and VirusTotal threat intelligence with the tech giant’s in-house developed Gemini AI model.

Powered by Gemini 1.5 Pro large language model, the company says it has significantly reduced the time required to reverse engineer a malware attack, with Google claiming that it took only 34 seconds for the large language model to analyse the code of the infamous WannaCry malware and identify a loophole that shuts it down. For the uninitiated, WannaCry was one of the biggest ransomware attacks the world has ever seen, affecting some of the largest organisations around the world.

Apart from reverse engineering malware attacks, Gemini can summarise threat reports into natural language so enterprises can easily decode and analyse how these attacks might impact their business. Google says Threat Intelligence has a large amount of network information which can be used to see a bigger picture in the cybersecurity world and help security experts understand what they need to prioritise. As it turns out, Google is also using Gemini to crawl the web to retrieve and categorise reports of online threats and offer ways to counter them.

Google plans to combine the information provided by Gemini with Mandiant’s human expertise with VirusTotal’s threat analysis to protect users from new malware attacks. The move comes after Microsoft unveiled the GPT-4 powered Copilot for Security, which is used by cybersecurity experts to ask questions about threats.

https://indianexpress.com/arti....cle/technology/artif