Malaysia is facing escalating AI-driven fraud attempts, driven by the surge in deepfakes.

Natalia Fritzen, AI Policy and Compliance Specialist at Sumsub, an identity verification and anti-fraud firm, told Fintech News Malaysia that her company observed a staggering 108% increase in Malaysia’s fraud rates between 2022 and 2023, with deepfakes soaring by 1,000% during the period.

“The number and quality of deepfakes is increasing and evolving daily worldwide,” Fritzen warned. “Even with the most progressive technology, it’s getting much harder to differentiate between a deepfake and reality.”

Using AI to combat AI

Fritzen discussed these issues during her presentation at the recent SC x SC Fintech Summit 2024 in Kuala Lumpur. In a panel discussion held on October 01 that featured AI and compliance experts, she highlighted the pivotal role of AI in reshaping compliance and risk management in the financial sector.

Fritzen emphasized Sumsub’s innovative AI-driven solutions, including liveness and deepfake detection tools, document forgery analysis, and face matching, to tackle rising fraud trends such as synthetic identity fraud. She highlighted her company’s commitment to developing proprietary AI technology that can keep pace with the increasing sophistication and prevalence of AI-driven fraud.

“The entirety of the products and services we provide have some type of AI technology behind them,” Fritzen said.

“We could talk about identity verification, which in our case is basically a liveness detection to make sure that you are the one behind the camera. We also analyze whether you are a deepfake or not, and whether you are being coerced or not. Simultaneously, to pass the authentication, you also have to submit your documents. We are going to make sure that there is no forgery. For that, we use AI to extract data, make sure the readability is good, etc. Those two things combine, we have a face match system that will detect whether the face in your liveness is the same in your documents. This is just one of the use cases.”

At the event, David Low, founder of Hata, a digital asset exchange and digital broker that received approval from the Securities Commission Malaysia (SC), acknowledged the need to rely on external expertise on non-core competencies.

As a startup, we don’t have the resources to build a team of 50 to 100 people, and building a talent pool takes a long time, so, we look at tech in the market, whether within Malaysia or outside, to integrate it within our processes, so that we can automate the identification of customer information, collect, verify and identify the risk scores of our customers, for us to monitor their transactions. All these are critical from a business perspective.

We are already using Sumsub, this helps us to scale up really quickly, without needing to hire a 100 people to do something that is fairly straight-forward but complex at the same time.

Sumsub’s systems can also cross-reference past applications to uncover repeat fraudsters. “Once we collected your data, and if you have activated that option, we can search your face among our pool of data and know whether you have already submitted an application to us and if you were rejected or labeled as a fraudster in the past,” Fritzen explained.

Sumsub also uses AI for transaction monitoring and to detect anomalies. In particular, its fraud network detection tool can map out coordinated fraudulent activities across individuals and entities, enabling detection of complex, multi-party fraud schemes.

“When you have different individuals or entities acting together using coordinated methodologies to commit fraud, if you have that option activated, we can try to map whether the fraud that is trying to be committed is connected to some others,” Fritzen said. “In that way, we can catch not only one fraud but the entire network.”

Customizable AI solutions

Fritzen highlighted the need for flexibility in AI systems to meet diverse client requirements. She noted that different clients have varying needs based on their specific compliance frameworks, particularly in areas such as anti-money laundering (AML) and counter financing of terrorism (CFT).

To address this, Sumsub is designing customizable AI solutions that allow clients to set parameters based on their unique compliance requirements.

Fritzen cited the example of a new rule generator the company recently launched. This tool enables clients to define specific rules that trigger additional verification steps based on suspicious activities. This rule can be, for example, flagging an applicant making repeated transactions to a sanctioned country for further review.

“We try to create solutions that are broad enough to satisfy anyone but also ensure that clients can customize these solutions for themselves,” Fritzen said. “Another cross-cutting feature that we try to have is to make sure that any AI solution that we develop is compliant and follows best practices.”

Navigating global AI regulations

Fritzen acknowledged the complexities of global AI regulations, emphasizing the challenges companies face in ensuring that their AI systems are compliant with a patchwork of regulations that vary by country.

“If you are starting to look at AI compliance, it’s critical to research which regulations apply to you based on which countries you operate in and also the industries that you operate in,” she advised. “The good news about industries is that, so far, most of the AI specific regulations already enacted tend to be sector agnostic.”

Fritzen emphasized the importance of understanding regulations like the European Union (EU)’s AI Act, which affects international companies, and stated that even Malaysian companies must pay attention to these regulations.

“The EU AI Act … is similar to GDPR, the data protection regulation … where even if you are based outside of the EU, if you process personal data from European residents, you should be compliant with GDPR. It has mechanisms in place to catch Malaysian companies or any other companies that are not EU-based,” Fritzen explained.

“What the AI Act says is that if a foreign company has an AI system or a general purpose AI model whose outputs can have an effect on the residents of the EU, then your technology should be compliant with this regulation.”

Meet the Sumsub team at the Singapore Fintech Festival (SFF) 2024

If you’d like to learn more about Sumsub’s AI capabilities in combating fraud, visit their booth at the upcoming Singapore Fintech Festival that is taking place from 06 to 08 November 2024. Speak to their team at Booth #3F17, Hall 3, or schedule a free demo with their experts via this link.

Featured image credit: edited from freepik