Arkose Labs Inc., a vendor of fraud prevention technology, uses AI to combat AI.
To take on the growing amount of fraudulent bots, the company is putting AI in cybersecurity, deploying constantly evolving machine learning models to block bots and try to shut down hackers.
Last year, the San Francisco company received a sizeable investment from PayPal, and it works with clients in industries such as travel, retail and gaming. It sells a fraud prevention service that it claims is ahead of Google's free reCAPTCHA both in terms of blocking automated spamming technologies and allowing real users through.
The company's success is partially due to AI, said CEO Kevin Gosschalk in an interview.
Arkose Labs employs global telemetry and a challenge test that Gosschalk said is effective at completely stopping or significantly slowing prospective fraudsters, yet that actual human users can easily pass with a first attempt pass rate of about 97%.
Essentially, Gosschalk said, Arkose can render 3D models in real time and ask the user to perform an action. The real-time generation makes it difficult for machine learning models to complete the action, which typically requires manual training to identify 3D models. The images switch quickly, as well; too fast to train a machine learning model on it, Gosschalk said.
Arkose put other AI in cybersecurity measures in place, as well, and while Gosschalk acknowledged that it's possible for non-human users to break through them, it can be too slow for fraudulent bots.
Kevin GosschalkCEO, Arkose Labs
"The core philosophy of our code is, how do we make it more expensive to attack than the value of breaking through?" he said.
Arkose Labs' own AI models learn with each attack, Gosschalk said, so that "every time we get attacked, we get better."
Constantly evolving and advancing is important for AI in cybersecurity and fraud prevention companies, such as Arkose, as hackers are quick to develop ways around security measures.
Cybersecurity AI circle
As Deloitte's "Tech Trends 2019" report notes, "Many of the cyberattacks that organizations confront today are orchestrated by Al engines on behalf of bad actors exploiting haps in their targets' security."
From this perspective, AI development is essentially stuck in a never-ending loop of updating and upgrading, as white hat agents and black hat agents try to outdo each other.
In some ways, it's analogous to the intensive use of antibacterial products. By using high volumes of antibacterial products in an effort to create a healthier world and better quality of life, humans have inadvertently created antibacterial resistant bacteria, or super bugs. These bugs can break through simple antibacterial defenses with ease, ultimately requiring more potent medicines.
Using fraud prevention AI isn't rare for organizations that employ AI and machine learning technologies, according to a late 2018 survey conducted by information-based analytics vendor Relx. About 33% of respondents said AI is helping to detect fraud, waste and abuse in their organizations. Organizations across a range of industries were interviewed.
For companies like Arkose Labs, using AI in cybersecurity is driving business and minimizing fraud and cyberattacks.