logo (23)

BforeAI

BforeAI is a pioneer in Predictive Attack Intelligence and Digital Risk Protection Services (DRPS). Our PreCrime™ platform uses behavioral AI to predict and automatically preempt malicious campaigns, making BforeAI the fastest, most accurate solution to stop attacks weeks before they happen.

Location:New York City, United States
Founded: 2020

Fraud Solution Profile

About BforeAI PreCrime™ Brand

BforeAI’s PreCrime is a preemptive scam, fraud, and impersonation protection solution that can prevent account takeover and credit card or credential stealing from customers’ customers, or protect our customers’ suppliers from being attacked in email compromises. This mitigates fraud, reputational harm risk from negative brand damage, or customer credential stealing. 

PreCrime Brand identifies a malicious infrastructure only minutes after its creation, puts a network disruption in place within minutes of identification, and requests action to various takedown operators, who subsequently disturb DNS resolution or content removal, resulting in infrastructure takedown. BforeAI’s privileged access to the takedown operators (including domain and DNS, content, industry alliances for abuse and malware prevention, law enforcement agencies, and independent response bodies) ensures malicious infrastructure is promptly removed. Our predictive technology is so effective that more than 80% of our takedowns are completed before there’s content on the infrastructure. 

BforeAI’s disruption partners, including VirusTotal, Quad9, Spamhaus, and Google Safe Browsing, subsequently put the identified malicious domain in a DNS resolution blocklist. Within 10 minutes on average, up to 75% of the traffic to the malicious infrastructure is already blocked.

 

A step-by-step breakdown of how BforeAI PreCrime works:

 

  • Data Collection and Pre-processing

PreCrime begins by collecting multiple network data points from thousands of sensors deployed across the Internet. The tool observes more than 1 billion infrastructures and 500million domains on a continuous basis – of them, 500,000 are created every day. Data is collected between 5 to 10 times per hour which enables observing any changes on a continuous and precise basis. This data is then preprocessed to remove noise and irrelevant information, ensuring that only meaningful interactions are considered. In total, PreCrime collects several terabytes of data on a daily basis.

  • Graph Construction and Feature Extraction

Next, PreCrime constructs a graph from the preprocessed data. Features such as query frequency, temporal patterns, and resolution paths are extracted and incorporated into the graph. Over 400 billion behaviors and edges are mapped in the graph database. These features provide a detailed view of domain interactions, which is crucial for accurate inference.

  • Application of Graph Inference Techniques

PreCrime applies various graph inference techniques to analyze the constructed graph. Four billion malicious behaviors are mapped in PreCrime.

Community detection algorithms identify clusters of domains that exhibit similar behavior, while anomaly detection algorithms highlight nodes with abnormal patterns. Link prediction algorithms are used to forecast potential future connections, helping to identify emerging threats.

  • Identification and Mitigation: Disruption and Takedown

PreCrime re-scores over 20 million suspicious infrastructures on a daily average out of which it predicts 100,000 future attack infrastructures. Based on the results of the graph inference analysis, PreCrime detects infrastructures that exhibit characteristics of malicious behavior. They are flagged for further investigation and mitigation. BforeAI’s preemptive approach ensures that potential threats are identified and neutralized before they can cause harm.

 

 Customers

Signify
Volksbank
Atlassian
Primark

 

 Industry
Ecommerce, Financial Services, Insurance, Telecom
 Primary Functionality
Fraud Platform
 Fraud Type
Account Takeover, Call Center Fraud, Content Abuse, Loyalty or Promo Abuse, Payment Fraud
 Technology
Machine Learning