On October 13, Coinbase reviewed the governance of crypto scams in Singapore, citing data from the local police that fraud related to cryptocurrencies in the first half of 2026 decreased by 43% year-on-year, from approximately S$114.9 million to about S$65.5 million. The company attributed this change to the cooperation between trading platforms, the police, and anti-scam centers, highlighting the role of on-chain analysis in quickly reaching out to potential victims. However, two things need to be distinguished: the decrease in losses is a statistical fact; whether the involvement of a particular platform contributed to that reduction cannot be strictly proven by the available information. Titling it " Coinbase Singapore's Scams Reduced by 43%" equates correlation with causation.
More specific actions can be verified from two official announcements by the Singapore Police Force. From mid-March to mid-April this year, the police's Anti-Fraud Centre and Cyber Investigations Department collaborated with multiple trading platforms such as Coinbase, Coinhako, Gemini, etc., using analysis tools like TRM Labs and Chainalysis to identify risky transactions. They directly contacted over 90 potential victims, preventing potential losses of more than S$2.86 million. Another round of joint operations in June further expanded coverage. In their latest review, Coinbase stated that the four rounds of joint operations involved a total of over 720 potential victims, avoiding potential losses of more than S$18.9 million. The latter figure is a summary of the cooperation's results and cannot be simply added to the figures from the police's first round, nor should it be interpreted as cash that has been recovered. Since the police operations involved multiple platforms and institutions, the credit for these successes should not be attributed solely to a single company.
The key to combating fraud is not just tracking addresses on the blockchain.
When the public thinks of blockchain anti-fraud measures, they often imagine that the core process is to identify the wallet addresses of suspects, after which the assets can be automatically frozen. However, the reality is much more complex. Fraud often begins on social media platforms, through fake job listings, or by impersonating government officials or exploiting emotional relationships. Victims may purchase crypto assets from legitimate exchanges only to then transfer them voluntarily to addresses controlled by scammers. Before a transaction is executed, the platforms see what appears to be an operation authorized by the victim themselves. To minimize losses, it is necessary to combine information about suspicious addresses, unusual transaction patterns, account behavior, reports received by the police, and timely human intervention. While the records on the blockchain have their transparent aspects, it is not possible to determine alone whether a person is being manipulated by fraudulent tactics.
The actions reported by the Singapore police exemplify this kind of collaboration. After law enforcement officers exchange information with the platforms, they intervene with potential victims via phone or in person. It's not about simply relying on algorithms to “block bad addresses”; instead, the risk detections are translated into actual blocking actions in the real world. The earlier the intervention, the greater the chance of obtaining an explanation before funds are irreversibly transferred. However, it is also necessary to avoid mistaking legitimate customers for victims, or allowing real scammers to impersonate the police to contact customers. Therefore, officials remind the public to verify notifications through credible channels and not to make further transfers due to so-called “urgent loss recovery” calls. Both technical identification and identity verification are essential.
“Potential loss” is the term most prone to errors in reports. It refers to the estimated amount that could be transferred if the fraud continues to progress; the actual amount lost, the amount later recovered, and the amount that would not have been transferred in the first place are all different indicators. The amounts prevented by the police and the cumulative figures of Coinbase are more suitable as clues to the scale of intervention, rather than being considered as recovered earnings in an accounting sense. The statistic showing a year-on-year decrease in losses in the first half of the year also needs to be considered within a broader context: victims’ reporting behavior, changes in fraud methods, other police measures, platform review strategies, market prices, and trading volumes can all collectively affect the results. Without comparative studies, it is impossible to accurately attribute the contributions of each party.
The scope of responsibility for trading platforms is expanding from account opening to before fund transfers.
This type of collaboration has practical significance for the Web3 industry. Trading platforms traditionally focus on identity verification and suspicious fund monitoring during account opening, while social engineering scams often occur with users who have genuine identities and legitimate accounts. A customer might pass through KYC completely, yet still transfer assets to an external wallet under the guidance of a scammer. If the platform only checks "whether it's the account holder," it may still miss "whether the account holder has been misled." This drives platforms to add more prompts, cooling-off periods, or manual confirmation before high-risk transfers, but also brings a trade-off between convenience and privacy. Excessive interception hinders normal payments, while insufficient intervention makes losses irrecoverable. Legal requirements also differ across jurisdictions; the Singapore model cannot be simply replicated globally.
For users, the most effective line of defense remains verification before making a transaction. Any request to bypass the regular process, urging a transfer on the grounds of confidentiality or urgency, or claiming to guarantee returns in blockchain investments should be seen as a red flag that warrants pause. Even if a call claims to be from an exchange or law enforcement agency, users should call back through the official website or known hotlines, and should not log in using links provided by the caller or enter verification codes. Users also need to understand that blockchain transfers generally do not have as easy a mechanism for reversal as bank transfers; therefore, the tactic of "transferring a small amount first to test" could also be the first step in a scam. These reminders are not meant to deny the value of crypto assets, but rather to acknowledge that the irreversibility of payment processes can increase the risk of social engineering attacks.
For policies and enterprises, the most important data to be made public in the future is not just a pleasing overall figure, but rather the accuracy of alerts, the time from discovery to contact, the proportion of victims who are successfully dissuaded, and the rate of false positives. Only if these indicators continue to improve can it be proven that the cooperative model remains effective even as it scales up. It is also necessary to consider what data the platform shares with the police, for how long, and how the use of that data is restricted, in order to prevent the unlimited expansion of personal information processing under the guise of security. The credibility of technical solutions comes not only from preventing fraud but also from providing an understandable level of protection for the rights of legitimate users.
Today, this news can be confidently reported as follows: In the first half of the year, losses from crypto scams in Singapore decreased by 43% year-on-year; multiple joint operations conducted by the police and platforms, involving Coinbase, have revealed specific results in identifying and contacting potential victims. This decline cannot be attributed solely to Coinbase, nor can the potential losses be referred to as funds that have already been recovered. For the industry, a more significant change is that crypto platforms are beginning to engage more systematically in risk intervention before transactions occur. This direction is valuable, but its true effectiveness still requires long-term, comparable, and transparent data verification.
Image caption: Photo of Brian Armstrong, the Chief Executive Officer, at a public event in 2017. This is not a photo from the anti-fraud operation in Singapore. Photography by Steve Jurvetson. Source: Wikimedia Commons. License notes can be found at https :// creativecommons.org / licenses / by /2.0/. The cover image is a cropped version of the original photo; for licensing details, please refer to https.
