东南亚各国政府和金融机构亟需找到一种安全且快速的数据共享方式，以遏制诈骗活动的激增。网络安全专家警告称，人工智能正使该地区的犯罪组织迅速扩大其活动范围。
Governments and financial institutions in Southeast Asia urgently need to find a way to safely and quickly share data if they hope to contain a surge in scams, as cybersecurity experts warn that artificial intelligence is enabling criminal operations in the region to expand their reach rapidly.
根据联合国7月发布的一份报告，2025年，犯罪组织在亚太地区从受害者手中窃取了高达1140亿美元，是两年前的三倍多，人工智能和其他技术的助长是主要原因之一。
Criminal enterprises stole as much as US$114 billion from victims across the Asia-Pacific in 2025, according to a United Nations report released in July, more than triple the amount from two years earlier, aided by AI and other technology.
随着“代理式人工智能”（agentic AI）的采用，诈骗和金融犯罪的祸害可能会进一步加剧。专家表示，这种技术可能使犯罪分子以前所未有的速度实现其业务的自动化和规模化。
The scourge of scams and financial crime could become worse with the adoption of agentic AI, which experts say could allow criminals to automate and scale their operations with unprecedented speed.
因此，该地区的监管机构和金融机构必须为最坏的情况做好准备，并建立有效的机制，以便安全地跨境共享数据，从而在诈骗活动扩散并造成损害之前识别并瓦解这些犯罪网络。
As such, regulators and institutions in the region need to prepare for a worst-case scenario and establish effective mechanisms to safely share data across borders, allowing them to identify and disrupt scam operations before they spread and do damage.
企业人工智能公司SymphonyAI的亚太区合规与金融犯罪行业顾问马丁·史密斯（Martin Smith）周二在吉隆坡的一场关于金融犯罪的圆桌讨论会上表示：“目前的问题不在于我们能否战胜犯罪组织，而在于我们会输得多惨。”
“At the moment it is not about whether we can win [against criminal enterprises]. It’s about how badly we lose,” Martin Smith, Asia-Pacific compliance and financial crime industry adviser with enterprise AI firm SymphonyAI, said in Kuala Lumpur on Tuesday at a panel discussion on financial crimes.
史密斯的这一评估紧随链上情报平台TRM Labs此前发出的警告。TRM Labs指出，诈骗团伙正大力投资训练代理式人工智能智能体，以便同时针对使用不同语言的多个受害者。
Smith’s assessment follows an earlier warning by on-chain intelligence platform TRM Labs, which pointed to scam syndicates investing heavily in training agentic AI agents to target multiple victims across different languages, all at the same time.
TRM Labs表示，完全训练好的人工智能智能体将消除对庞大劳动力的需求，并实现完全远程运营，从而使诈骗团伙能够指数级地扩大其规模和覆盖范围，同时变得更加难以被抓获。
Fully trained AI agents will remove the need for a large workforce and enable fully remote operations, allowing syndicates to exponentially increase their scale and reach while becoming more difficult to catch, according to TRM Labs.
史密斯表示，该地区的监管机构和机构继续落后于犯罪企业，因为它们被困在自己的孤岛中，受到跨境数据共享笨拙规则的阻碍，难以整理信息以帮助清除犯罪活动。
Smith said regulators and institutions in the region continued to lag behind criminal enterprises as they were stuck operating in their silos, hindered by unwieldy rules on cross-border data sharing that made it difficult to collate information to help weed out criminal activity.
马来西亚透明国际主席雷蒙德·拉姆表示，各国政府、执法机构和金融机构需要扩大合作，包括交换有助于查明犯罪资金踪迹的金融数据。
Raymond Ram, president of Transparency International Malaysia, said governments, enforcement agencies and financial institutions needed to expand their cooperation to include exchange of financial data that could help pinpoint criminal money trails.
“我们不能仅仅依赖[司法互助]。我们需要金融当局之间更多的数据共享，”拉姆说。
“We cannot depend just on [mutual legal help]. We need more data sharing between financial authorities,” Ram said.
犯罪学家沙米尔·拉贾杜赖（Shamir Rajadurai）表示，制度应对至关重要，因为个人没有机会对抗针对他们资金的大规模犯罪企业。
Shamir Rajadurai, a criminologist, said an institutional response was vital as individuals would stand no chance against the large scale of criminal enterprises targeting their money.
“这些犯罪分子是在组织层面活动的。因此，我们需要组织对组织，以毒攻毒，”沙米尔说。
“These criminals are operating at an organisational level. So we need to get organisation versus organisation, fight fire with fire,” Shamir said.