Cloudflare announces AI-powered Fraud Detection tool

[harlan4096]

Cloudflare announced a new AI-powered fraud detection tool this week that is designed to protect businesses and consumers against various types of fraud.
Cloudflare Fraud Detection protects against fake account logins, account takeovers and stolen credit cards. The new feature can be "deployed in seconds to any website on the Cloudflare network" according to Cloudflare's introductory blog post.

Cloudflare customers may choose behaviors for every type of fraud that the new AI-powered tool may detect. Customers may block fraudulent traffic outright or react to it in another way, for example, by redirecting high risk users to "an alternate online experience with fewer capabilities".

Fraud on the Internet has exploded in the past couple of years according to Cloudflare. The Federal Trade Commission reported that consumers lost US $8.8 billion in fraud in 2022, which is a 400% increase over the money lost in 2019.

The Covid-19 pandemic, or more precisely the reactions to the pandemic, has moved people online at a faster speed than ever before. With that rise came also an increase in fraud.

Organizations have two main ways when it comes to fraud, according to Cloudflare: either implement protective features, such as requiring two-factor authentication or blocking access from known VPN connections or regions associated with high numbers of charge backs, and risk alienating users and losing business, or "bear the cost of fraud", and deal with higher infrastructure costs and less efficient employees.

Cloudflare Fraud Detection is powered by a machine learning model that relies on a large, global network; this enables Cloudflare to process data in 0.2 milliseconds and run its model in real-time and, according to the company, without performance impact.

The tool is designed to address four major fraud types, but without hurting the user experience:

1. Fake Account Creation -- Fake accounts are used by bad actors for a variety of purposes: from gaining access to promotional rewards several times to spamming or trying to influence elections using an army of bot accounts.
   
2. Account Takeover -- Using phishing, hacking and other attack forms to gain access to the accounts of users and employees of a service.
   
3. Card Testing and Fraudulent Transactions -- Testing the validity of stolen credit cards and using these cards to make online transactions.
   
4. Expediting -- Circumventing the normal user flow to obtain "limited availability goods or services".
   

Cloudflare provides details on two of the fraud types in the blog post, and promises to publish information on the remaining two in a follow-up. The company claims that its Cloudflare One product is already detecting 72% of fake account creations, and that it hopes to bring the number close to 100% using the new fraud detection tool.

Cloudflare plans to launch Fraud Detection in the second half of 2023. Interested customers may sign-up to participate in the initial early access phase already. Customers may opt-in to provide training data sets to Cloudflare to improve the product on specific use cases.
...

Continue Reading