Ant Group launches deepfake detection challenge

The deepfake detection boom has taken a financial turn in China, with the announcement of the Global Multimedia Deepfake Detection Challenge 2024, a contest sponsored by the Inclusion Conference on the Bund and digital conglomerate Ant Group (of which Alibaba is majority owner).

A news release promises a total prize pool of 1,000,000 Chinese yuan (around US$137,000) to be won by entrants that can innovate, develop and test deepfake detection models that are more accurate and effective in a series of real-world scenarios. Registration closed on July 31 and testing is currently in progress. Results are to be announced in early September.

Per the release, the deepfake challenge is structured into two tracks: an image track and an audio/video track, featuring comprehensive datasets with “multi-dimensional face forgery methods.” The twenty models that are best at predicting the probability of a deepfake will advance to the final round, where a panel of experts will evaluate their deepfake detection tech against criteria including “accuracy, generalizability, innovation, practical applicability and the interpretability of their solutions.”

Top three teams from each track will receive invitations to a special ceremony at the Inclusion Conference on the Bund in Shanghai.

Zoloz, a subsidiary of Ant Group that provides biometric identity verification – and, as of recently, deepfake detection software – is listed as contributing technical support. Other companies in the Ant Group ecosystem, Ant Security Lab and Alibaba Cloud, are also on board, as is the Advanced Technology Exploration Community (ATEC) and the Cybersecurity College of the University of Science and Technology of China.

Generative AI fueling both sides of the deepfake fraud war: Experian report

While the risk presented by generative AI, deepfakes and biometric fraud have been well established, Experian’s annual Identity and Fraud Report provides additional insights into current trends in identity fraud.

Among noteworthy findings are the correlation between the key element of trust and the ability to repeatedly identify customers with accuracy and ease. Sixty-three percent of respondents say it’s extremely or very important for businesses to be able to recognize them online. In short, solid identity verification adds legitimacy and trust.

At present, financial services companies are the most trusted, with retail banks, P2P lending and “buy now, pay later” financing all listed as top trusted organizations by U.S. consumers.

A fascinating relationship is evolving between the two heads of generative AI: Gen AI as a tool driving the plague of deepfake fraud and Gen AI as the ultimate defense against this same fraud. Experian’s data shows companies reporting high engagement and investment in Gen AI-based security defenses. On the other hand, 70 percent of businesses say AI fraud is expected to be a top-two challenge for them.

The two-headed monster will continue to grow, but right now, investment in Gen AI fraud detection and prevention is low, with businesses still allocating most of their potential funding in preventing legacy fraud techniques like identity theft and first-party fraud.

Indeed, most of the top five fears reported among U.S. consumers are described in relatively low-tech terms: identity theft (84 percent), stolen credit card information (80 percent), online privacy (67 percent), phishing emails or phone scams (65 percent), and false information or fake news and ads (49 percent).

“With digital transactions increasing every day and new technology changing the fraud landscape, our latest report underscores the need for businesses to review their current strategies and invest in the right tools to address the evolving complexity of fraud schemes of the future,” says Kathleen Peters, chief innovation officer for Experian in North America. “Companies need to take a multilayered approach to fraud prevention that leverages the right data, analytics and technology in an orchestrated way to combat fraud and build trust and positive experiences with legitimate customers.”

Customers trust physical biometrics; companies need to get on board

Experian’s results show that consumers have greater trust in physical biometrics and behavioral analytics than in legacy cybersecurity methods like passwords and multifactor authentication (MFA). However, “less than 30 percent of businesses are using these solutions – showing companies could consider investments in physical and behavioral analytics to verify identities and combat fraud.”

Experian, a Dublin-based data analytics and consumer credit reporting company, recently acquired behavioral analytics firm NeuroID.

The full Identity and Fraud Report and methodology are available on Experian’s website.

Article Topics

Ant Group  |  biometrics  |  competition  |  deepfake detection  |  Experian  |  fraud prevention  |  Zoloz