YC-backed ‘Rulebase’ Builds AI-driven Voice Fraud Detection System

On November 6, 2024, Y Combinator took to its LinkedIn to announce the launch of an AI-driven fraud detection model by Rulebase. According to the post, this model integrates with IVR and CCaaS systems to prevent account takeovers, vendor fraud and voice phishing. 

Rulebase said on YC’s official blog, “We use AI to analyse voice biometrics and call behaviour, blocking account takeovers, impersonations, and voice clones.”

With rising AI-enabled voice scams, it has become difficult to tell if the person on the other end of a phone call is a friend or a fraud. This is exactly the challenge this company claims to be working on. It runs voice phishing simulations to train employees against bad actors. 

According to a McAfee survey conducted last year, 70% of people said they weren’t confident that they could tell the difference between a real and cloned voice.

YC recently released a press release stating that the startup creates voice agents that simulate phishing attacks on employees, helping them strengthen defences while identifying vulnerabilities. With this, the startup seems to be positioning itself to join the ecosystem of tech companies fighting deepfakes. 

Rulebase was founded by experts Gideon Ebose and Chidi Williams, who bring their experience from building defensive systems at Goldman Sachs, Microsoft, and other startups. They aim to tailor Rulebase’s mission to the needs of every business.

As reported by AIM last year, deepfakes are fake voices, images or videos that use AI to swap one person’s identity with another’s. These are made with advanced machine learning tools like autoencoders or GANs, making it easier to create realistic-looking content that can trick people.

According to Y Combinator’s blog, voice phishing and deepfake voices increased by 700% in 2023, which has further significantly risen in 2024. Fraudulent calls increase average handling time (AHT), putting extra pressure on service teams and reducing support for real customers. Voice phishing can also cost up to $14 million per incident, draining resources and affecting the bottom line.