Perception Point, an internet security platform, has introduced an innovative AI model to combat the growing threat of AI-generated email attacks. By utilizing large language models (LLMs) and deep learning architecture, the company’s detection technology aims to identify and prevent business email compromise (BEC) attacks that are facilitated by generative AI technologies.
With cybercriminals increasingly using generative AI technology to carry out sophisticated and precisely targeted attacks, organizations of all sizes are at risk. These attacks, particularly in social engineering and BEC, leverage generative AI to create personalized and human-like emails.
According to Verizon’s recent data breach investigation report, over 50% of social engineering incidents are attributed to BEC. Perception Point’s own annual report for 2023 highlights an 83% increase in BEC attempts.
To address this escalating threat, Perception Point has developed a detection model based on LLMs that incorporate transformers. These AI models can comprehend the semantic context of the text, similar to renowned LLMs like OpenAI’s ChatGPT and Google’s Bard. By recognizing distinct patterns in LLM-generated text, Perception Point’s solution can effectively identify and thwart gen AI-based threats.
The company argues that conventional security vendors often fall short in achieving the required level of detection accuracy through contextual and behavioral analysis. While advanced email security systems employ contextual and behavioral detection, they struggle to identify the newly enhanced attacks facilitated by generative AI. These attacks bypass the usual patterns that detection methods were originally designed to recognize.
Furthermore, existing solutions in the market rely solely on post-delivery detection, allowing malicious emails to sit in users’ inboxes for extended periods before being removed.
Perception Point stands out by offering a comprehensive scanning of all emails, quarantining those identified as malicious before they reach the user’s inbox. This proactive approach eliminates the risks and potential damages associated with detection-based methods that only address threats after they have infiltrated the system.
In addition to its detection model, Perception Point provides a managed incident response service, relieving customers’ SOC teams from the responsibility of swiftly responding to incidents and deploying new algorithms in real time to counter emerging threats.
Perception Point claims its model excels in processing incoming emails, with an average time of 0.06 seconds. Originally trained on hundreds of thousands of malicious samples, the model is continuously updated with new data to optimize its effectiveness.
Tal Zamir, CTO of Perception Point, highlights that attackers are exploiting fake emails to impersonate trusted organizations, deceiving employees into transferring money or disclosing confidential data. Employees are often considered the weakest link in an organization’s security, making them susceptible to BEC text-based attacks that bypass traditional email security systems and land in users’ inboxes.
The emergence of generative AI has significantly boosted impersonation, phishing, and BEC attacks, empowering cybercriminals to operate with greater speed and scale. High-quality attacks, which previously required extensive time and effort, can now be accomplished within minutes using generative AI.
To minimize false positives resulting from legitimate emails that heavily rely on generative AI, Perception Point’s model incorporates a unique three-phase architecture. After an initial scoring process, the model utilizes transformers and clustering algorithms to categorize email content. By integrating insights from these stages with additional data, such as sender reputation and authentication protocol information, the model predicts whether an email is AI-generated and determines its potential threat.
Zamir mentions that his company aims to develop AI capabilities that can sift through vast amounts of data, identifying potential threats, and providing customers with actionable intelligence. He emphasizes the need for AI security solutions to prevent, detect, and respond to the increasing complexity of the threat landscape.
Perception Point acknowledges the new avenues for potential attacks created by the integration of generative AI bots into collaboration apps, browsers, and cloud storage services. The company is committed to ensuring its clients can leverage the power of generative AI without compromising their security posture.
