SG develops new tool to verify digital content and tackle deepfakes

share on

twitter /
facebook /
linkedin /
- email
- telegram
- whatsapp
- wechat
- pinterest
- line
- snapchat
- reddit

Singapore’s Centre for Advanced Technologies in Online Safety (CATOS) will introduce a free online tool in 2026 to help users verify whether images and videos have been altered, according to a report by The Straits Times (ST).

The tool, Provo, reportedly allows publishers, brands, and creators to embed provenance labels into visual content. Users can view these labels through a free browser extension. According to ST, the labels function as “nutrition tags”, carrying information such as the original publisher, posting date, and edits made before the content went live.

The provenance markers are reportedly removed automatically if a visual asset is manipulated by an unauthorised party. The traceability also remains intact even when content is reposted or shared across platforms.

Don't miss: Can an 'AI generated' label really build trust, or will it slow the industry down?

Speaking to ST, CATOS director Yang Yinping said users should no longer rely on telltale signs such as unnatural blinking or mouth blurring to spot deepfakes. Improvements in AI-generation techniques have made surface-level cues unreliable, she explained. Yang said provenance tools offer a necessary complement to deepfake detection, which she described as a “cat-and-mouse game”.

The report added that Adobe and CATOS began collaborating in 2024 to develop provenance technologies in Singapore. More details on Provo’s public release are expected in May 2026, CATOS’ head of systems engineering Therese Quieta told ST.

Quieta added that the team is working on a secure identity verification process to prevent impersonation attempts, with Singpass integration under consideration. CATOS is also in discussions with publishers and creators to build a wider trust ecosystem around the tool.

Alongside Provo, CATOS has developed a deepfake recognition platform called Sleuth, according to ST. The tool is currently used by employees in some public agencies and media companies. Sleuth reportedly analyses pixel-level anomalies to determine whether faces, audio, or environments have been manipulated.

CATOS principal scientist Soumik Mondal told ST the team is still assessing whether Sleuth can be made available to the wider public, citing accuracy thresholds and infrastructure demands.

MARKETING-INTERACTIVE has reached out to CATOS for more information.

The launch of Provo builds on CATOS’ earlier initiatives. The centre was officially launched in May 2024 by Minister for Communications and Information Josephine Teo at the inaugural Online Trust & Safety Forum, with a total of SG$50 million in funding under the Research, Innovation and Enterprise 2025 Plan.

Hosted by A*STAR, CATOS was set up to strengthen Singapore’s ability to address online harms such as misinformation, hate speech, and AI-generated content. The funding supports new detection capabilities, continued research through open calls and innovation challenges, and a sandbox environment to test emerging tools.

CATOS also signed an MoU with Adobe in 2024 to explore content provenance technologies in Singapore, and works with community partners such as CheckMate to enhance fact-checking efficiency. The inaugural forum gathered more than 250 experts to discuss trends in online trust and safety.

share on