How Grab is using technology to improve trust and safety

Southeast Asian unicorn Grab is tapping artificial intelligence and other technologies to keep its users safe and cyber criminals at bay

As Southeast Asia’s largest super app that connects millions of consumers to millions of drivers, merchants, and businesses, Grab has had to deal with the challenges of ensuring the safety of its users while deterring fraudsters from profiteering from its platform.

To address the challenges, Wui Ngiap Foo, Grab’s head of integrity, an area that spans user trust, identity and access management, safety, product security, and financial risk, said the company has designed a plethora of features in its app, starting with facial recognition checks before a ride begins.

Speaking at a recent media briefing, Foo said Grab drivers are required to take selfies, which are matched against their registration records. This helps to deter unregistered drivers from using another driver’s account, and additional selfies may be required to deter drivers from passing their phones to another driver.

The Grab driver’s app will also conduct a “liveliness check” on selfies by requiring drivers to perform gestures such as nodding, so using a photo of the driver in place of a selfie would not pass the test.

Foo said the selfie feature has been successful in preventing unregistered drivers from coming onboard, renting of driver accounts to groups of drivers sharing the same vehicle and even the sale of driver accounts on the black market.

Grab has also rolled out the selfie feature for passengers in Malaysia, where it is legally required to verify the identity of passengers. It plans to introduce the same feature in Singapore, Thailand, Indonesia and Vietnam.

But as Grab’s passenger base is much larger than that of its drivers, selfies are only required if a user has not logged on to the service for 90 days. This has deterred bad actors such as drug peddlers from using Grab for illegal activities.

User experience

Amid the Covid-19 pandemic, where mask wearing has become the norm across Southeast Asia, these measures could add more friction to the user experience, since drivers and passengers would have to remove their masks and put them on again after the verification process is completed.

This challenge was posed to Grab’s data science team which has since developed various artificial intelligence (AI) models that can now perform facial recognition in different lighting conditions, even if a person has a mask on, with an accuracy rate of 99.5%.

The company is now planning to enhance its facial recognition feature, by interpreting how light reflects off a person’s facial features to ascertain if the person is real without requiring gestures, Foo said.

Grab has also applied its AI capabilities in its messaging system that facilitates communication between drivers and passengers. Using natural language processing, it has been able to detect and filter out abusive language, as well as phishing, fraud-related and sexual messages, Foo said.

During a ride, trip monitoring technology is used to track deviations from the planned route, as well as unplanned stops that will trigger a notification on the passenger’s app to check if he or she is fine.

Using telemetric data, the technology is being improved to detect crashes which could trigger an emergency call for an ambulance to be despatched. “Or, if a ride terminated too fast or too late, we want to know what’s going on and make sure the driver and passenger are safe,” he said.

Grab also sends drivers safety reports to educate them about their driving behaviour in a bid to improve service quality. “Rides with higher frequencies of harsh braking, cornering or speeding usually receive a lower rating from passengers. We feed those insights to drivers so they can get better,” said Foo.

Read more about IT in ASEAN

But in developing markets like Indonesia, Thailand and Vietnam, Grab’s fleet of motorcycles are more often used by commuters, prompting the company to invest heavily in telemetry for two-wheelers, which is not well-researched.

“We have access to massive amounts of telemetric data, but bikes take small roads, go up kerbs, and swerve onto storefronts, so trying to predict driving risk is not easy.”

Besides facilitating services in the physical world, Grab has started offering digital services, notably financial services including loans, mobile payments, insurance and investments, making it a target for cyber criminals who have used social engineering and phishing websites to trick Grab users into revealing their passwords.

In summing up Grab’s approach in protecting users, Foo said the company, which has worked closely with law enforcement agencies to bust criminal syndicates, not only invests in technology, but also in user awareness and education.

“We invest a ton in AI and machine learning, because we believe that’s the only way to keep up with the sheer volume of attacks and fraud,” said Foo. “We’re betting very aggressively on the fact that AI is going to be one of our major tools to combat safety incidents, both online and offline.”

Read more on IT risk management

Data Center
Data Management