TL;DR
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Bottomline risk and fraud officer Katie Elliott told PYMNTS that AI is helping fraudsters make B2B payment attempts faster and broader, while manual verification remains costly for finance teams. She argues that payment controls should combine multiple signals, automate routine decisions and send unusual requests to people for review.
AI is allowing fraudsters to send more B2B payment scams while companies face the expense of checking each request, Bottomline senior risk and fraud officer Katie Elliott told PYMNTS. Her account points to a growing challenge for finance teams: verify suppliers and payment instructions quickly enough to keep funds from moving to impostors, without routing every transaction for manual review.
Elliott said scammers can use available AI tools to make attempts “bigger, broader, faster” than more targeted schemes of the past. She described a shift toward mass phishing and spam, which increases the volume of suspicious requests companies may need to assess. The interview does not quantify how much fraud has increased or provide a measured comparison of AI-assisted and non-AI attacks.
The timing of a payment can make prevention more important than recovery. Elliott said authorized payments may move “in an instant,” and that fraudsters try to remove funds as soon as they receive them. That makes checking the counterparty and payment instruction before sending a stronger control than relying on recovery after a mistaken transfer.
She recommended using multiple signals rather than relying on a single data point. Those may include digital identity, phone information, email-domain history and whether a request fits the supplier’s established payment behavior. Elliott said buying access to third-party data and verification tools can be expensive, particularly for smaller businesses, and suggested payment networks may be able to spread those costs across more transactions.
Why Payment Checks Must Scale
The issue is a mismatch in operating costs. A fraudster can make many attempts and needs only some to succeed, while a finance department must protect legitimate payments without creating delays or hiring a large team to inspect every request. If AI increases the number of attempts, manual checks alone may become harder to sustain.
That makes verification before payment an operational question, not just a fraud team concern. Companies must balance the expense of stronger checks against the consequences of sending money to the wrong recipient. Faster transfers can narrow the time available to respond once a payment is authorized.
The proposal to share verification capabilities through payment networks is a claim about potential value, not a demonstrated outcome in the source material. If networks can make useful data and tools available at lower cost, smaller firms may be better able to assess payment requests. The report gives no pricing, performance results or evidence that shared services prevent fraud in every case.
Products Worth Considering
Dell XPS 13 9350 Copilot+ Laptop, 13.4-inch 120Hz Display, Intel Core Ultra 7 256V, 16GB RAM, 512GB SSD, Ubuntu Linux 24.04 LTS
Apple Mac mini Desktop Computer with M4 chip, 10‑core CPU and GPU, 24GB Unified Memory, 512GB SSD, Gigabit Ethernet
Dell UltraSharp 27 Thunderbolt Hub Monitor (U2726DER)
As an affiliate, we earn on qualifying purchases.
From Onboarding to Ongoing Checks
Supplier onboarding has traditionally helped establish who a business is and where it should be paid. Elliott’s concern is that AI-generated identities could weaken confidence in identity checks performed at one point in time. The interview does not describe a specific case of an AI-created supplier identity or establish how often such cases occur; it identifies identity generation as a risk she is watching.
Elliott’s suggested approach combines automation with human review of exceptions. Routine activity can be handled automatically, while changes to payment instructions, unusually high transaction volume or an abnormal payment request can be escalated. She gave the example of a supplier that normally receives a steady payment suddenly asking for three times the usual amount; such a change, she said, should go to a person for approval.
The discussion appeared in PYMNTS’ coverage of its B2B Payments Event 2026. It presents Elliott’s assessment and recommendations; it is not a survey of finance departments or an independently verified measurement of fraud trends.
“They are using the technology that’s out there, the AI, every tool available to them in order to make their attempts bigger, broader, faster.”
— Katie Elliott, senior risk and fraud officer at Bottomline, speaking to PYMNTS
How Much Risk Is Measured
The source material does not provide figures for B2B fraud losses, the share of attacks involving AI, or the frequency of supplier impersonation. It also does not compare the cost or effectiveness of manual checks, automated systems and network-based verification. Elliott’s account is an expert’s assessment in an interview, rather than an independently documented industry-wide finding.
It remains unclear which verification services payment networks currently offer, what they cost, how broadly businesses use them, and how well they identify fabricated identities or compromised supplier accounts. The interview also does not specify how companies should set escalation thresholds or resolve cases where available signals conflict.
Controls Move Before Payment
For finance teams, the practical next step described by Elliott is to examine which payment requests can be checked against several signals and which changes should trigger human approval. Examples include altered bank details, spikes in payment activity and amounts that differ sharply from a supplier’s normal pattern. The source does not announce a new product, policy or deadline for adopting these controls.
Payment networks’ potential role in sharing verification tools remains a question for providers and business customers. Any assessment of that approach would need clearer information on cost, coverage and performance. PYMNTS said readers could watch its full interview with Elliott for further discussion; the source material does not identify a subsequent announcement or a scheduled milestone.
Key Questions
What did Katie Elliott say AI is changing about B2B fraud?
Elliott said AI tools let fraudsters make attempts “bigger, broader, faster.” The source gives her assessment but no data measuring the overall increase in attacks.
Why does payment speed matter?
Elliott said an authorized payment can move quickly and fraudsters may remove funds soon after receiving them. That makes checking the recipient and payment instruction before sending important.
What information can companies use to verify a payment request?
Elliott cited digital identity, phone information, email-domain history and behavioral patterns. She advised against relying on only one signal when deciding where to send money.
Does Elliott recommend sending every unusual payment to manual review?
She described automation for routine activity and human review when patterns break, including changes to payment instructions, unusually high payment volume or an amount well outside a supplier’s normal range.
What is still unknown about network-based verification?
The interview does not provide prices, adoption figures or measured results for verification offered through payment networks. Its potential to lower costs for businesses, especially smaller firms, remains a proposition rather than a quantified finding.
Source: rss
