Agentic AI Fraud Is Turning Scam Prevention Into An Operations Problem

Fintech News Singapore reported findings from BioCatch's 2026 AI, fraud, and financial-crime research, which surveyed 1,440 fraud, AML, risk, and compliance professionals at financial institutions across 25 countries. The useful detail is not only that AI is being used in scams. It is that fraud professionals now expect more autonomous systems to appear in the fraud workflow.

According to BioCatch, 80 percent of surveyed respondents said their institution had already encountered attacks using agentic AI. The same release said 84 percent see AI agents as the industry's greatest exploitable vulnerability in the next year, 88 percent said AI has already increased fraud sophistication, and 72 percent believe it will be very difficult to distinguish legitimate AI-assisted activity from malicious or manipulated AI activity when AI agents commonly initiate transactions.

The tension is clear: banks and fraud teams are trying to use AI agents defensively while criminals may use similar automation to scale manipulation, phishing, credential attacks, transaction fraud, and social engineering.

For Southeast Asia, the practical relevance is operational. The affected surfaces are not limited to banks. If scam workflows become faster and more automated, the pressure can land on any business that handles account access, customer identity, refunds, payouts, transactions, or support conversations.

The APAC signal in the source is especially relevant. Fintech News Singapore cites the BioCatch research showing that financial institutions in APAC were among the regions reporting heavy exposure to agentic-AI attacks, with 79 percent saying their institution had encountered attacks using agentic AI. It also reported that 81 percent of APAC respondents saw attempted fraud increasing year over year.

Those numbers do not prove the same impact in every Southeast Asian market, but they show why financial institutions and operators in the region should pay attention. The practical question is no longer whether scams use AI. The question is whether companies can notice suspicious behavior fast enough, explain their decisions clearly enough, and protect customers without making legitimate users feel locked out.

The operator takeaway is that fraud prevention is becoming a live workflow problem.

If you run a team, the lesson is not to buy every new security product. Start by mapping the points where trust is created or lost:

- account signup
- login and password reset
- payment approval
- refund requests
- support chats
- identity checks
- unusual behavior after account access changes
- high-risk customer instructions

Then ask which step is still too manual, too slow, or too easy to spoof.

For a bank, that may mean stronger behavioral detection and better scam-intervention flows. For an e-commerce company, it may mean better refund and account-takeover checks. For a SaaS company, it may mean tighter admin-action review. For a small business, it may mean clearer rules for payment changes, vendor bank details, and customer support escalation.

For workers, the preparation is practical. The useful skill is not only knowing AI prompts. It is knowing how fraud, verification, customer support, payments, and approvals actually move through a business.

If your work touches customers, money, accounts, or data, this points to a likely direction: more verification, more escalation rules, and more need for judgment when something looks wrong. Some checks may be automated. Some will still need humans who can spot suspicious patterns, write clear handoff rules, and explain why a decision was made.

For individuals, the same lesson applies at a personal level. Treat urgent payment changes, login links, investment offers, hiring messages, and customer-service instructions with more caution. If AI can make scams more convincing, the defense is slower verification, not faster reaction.

Watch how Southeast Asian banks, wallets, payment providers, and digital platforms respond in three areas.

First, customer verification. Companies will need better ways to confirm whether a user is real without making every interaction painful.

Second, explainability. BioCatch said 72 percent of surveyed leaders expect it to be very difficult to distinguish legitimate AI-assisted actions from malicious or manipulated AI activity in a future where AI agents commonly initiate transactions. That matters because fraud decisions affect real customers. If account access is denied or a payment is stopped, teams need to explain why.

Third, scam education. Fintech News Singapore reported that surveyed professionals saw AI increasing fraud sophistication through deepfake-enabled social engineering, automated phishing, and automated money laundering or transaction fraud. That suggests the next wave of fraud prevention will combine technology, behavior analysis, and public awareness.

The bigger signal: as AI agents enter both fraud and defense, trust becomes an operating capability. Companies that handle payments, accounts, support, or sensitive customer requests will need stronger workflows, not just stronger warnings.

- The Rise of Agentic AI in Cyber Fraud and Scams - Fintech News Singapore
- World Banking Leaders See AI Agents as Industry's Greatest Vulnerability in Next Year - BioCatch
- The Future of Digital Trust - BioCatch