Radar’s broadened coverage now spans payment methods, processors, and platforms

Stripe’s fraud prevention product, Radar, has expanded well beyond its traditional scope of card payments made directly on Stripe. The updates, announced at Stripe Sessions, extend protection across all supported global payment methods, add new signals for transactions processed elsewhere, and introduce custom fraud models for businesses with complex risk profiles.

Fraud detection now applies uniformly across all payment methods

Radar previously focused its fraud detection on card transactions. Now, it protects volume across bank debits, buy now, pay later (BNPL) options, crypto, digital wallets, real-time payments, and cash vouchers. Crucially, the intelligence is shared across these methods: if Radar detects and blocks a fraudulent pattern from a stolen card, the associated IP address and device fingerprint are flagged for all other transaction types on the network. Stripe reports that businesses using Affirm, Cash App, Klarna, and PayPal saw a 71% reduction in suspected fraud over a five-month period.

New risk signals for off-Stripe transactions

Businesses that use multiple payment processors can pull Radar’s risk signals to supplement their own fraud models. Two new signals are now available for these off-Stripe transactions. The first predicts whether a payment is likely to trigger an early fraud warning from the card network, allowing merchants to proactively refund the transaction and protect their dispute rate. The second predicts the likelihood that a payment will result in a fraudulent dispute, which can inform decisions on refunds, evidence gathering, or dispute strategy.

Custom fraud models for complex risk profiles

For businesses whose risk factors don’t fit a standard profile, Radar now supports custom fraud models. Merchants can pass their own structured data—product catalog details, loyalty status, behavioral metrics, or any other relevant metadata—to Stripe. This is combined with Stripe’s network data to create a model built for that specific business. Early adopters of custom models report detecting at least 15% more fraud without an increase in false positives.

New defenses target compute and token abuse

Fraud isn’t limited to stolen money; it also involves stolen compute. Bad actors cycle through free trials, create multiple accounts, and rack up usage-based charges they never intend to pay. Radar now has tools to address several of these patterns.

Blocking multi-account abuse

Multi-account abuse occurs when one actor creates several accounts to reuse promotions or distribute stolen card activity and avoid detection. This is a significant problem in the AI space, where more than one in six sign-ups are linked to this type of abuse. Radar now evaluates each new account in real time, using device fingerprints, IP addresses, email domains, and historical abuse patterns from across the Stripe network to block suspicious accounts before they can cause harm. This capability works whether the account is managed on or off Stripe. ElevenLabs reports that it has used this feature to block 2,000 users a day from abusing its free tier.

Predicting pay-as-you-go abuse

Consumption-based pricing creates a window for abuse: customers can consume resources throughout the billing cycle only to default when the invoice arrives. Radar can now predict nonpayment as usage accumulates, allowing businesses to intervene before the customer is billed. This could involve requiring a top-up, cutting off service, or taking another action aligned with the business’s risk tolerance.

Scoring bot-driven payments

With the rise of agentic commerce, Radar has introduced a bot score for payments made on Stripe Checkout. This score evaluates the likelihood that a payment was made by a malicious bot rather than a legitimate agent acting on a customer’s behalf. The score can be used to enforce anti-scripting policies, such as blocking automated purchases of limited-availability items or flagging high-velocity orders for manual review.

Platform tools for evaluating merchant risk

To help platforms manage the risk of fraudulent merchants, Radar is adding new features that operate on and off Stripe. Platforms now have access to 0-to-100 fraud scores for every business and transaction, AI-powered insights explaining why accounts were flagged, account history and note-taking tools, and account-level metrics for disputes, declines, refunds, and payments.

Three specific signals have also been introduced to help platforms monitor merchant risk:

  • Fraudulent website: This signal analyzes a business’s website the way a human fraud analyst would, looking for indicators like unrealistically low prices, AI-generated copy, misspelled brand URLs, or other signs of a fraudulent site. It can be used during onboarding to automate verifications or flag accounts for manual review.
  • Fraudulent merchant: This signal identifies whether a new or existing account poses a fraud risk, using bank account information, business details, transaction activity, and dispute patterns from across the Stripe network. Platforms can respond by raising a review, pausing payouts or payments, rejecting the account, setting reserves, or requesting identity verification.
  • Merchant delinquency risk: This signal predicts whether a business is at risk of accruing a negative balance that stays negative for 60 days or more. Platforms can use the prediction to proactively adjust payout schedules, require reserves, or flag merchants for closer review.

Smart Disputes gains evidence recommendations and an evidence library

Smart Disputes, Stripe’s AI-powered dispute management product, now offers more targeted strategies for winning disputes. It can analyze each dispute and provide AI-powered recommendations for specific evidence fields, such as tracking numbers or customer usage logs. According to Stripe, businesses that add this AI-recommended evidence win disputes three times more often than those that add no evidence.

The manual effort of submitting evidence is also being reduced through the new evidence library. Businesses can upload and store standard documents—terms and conditions, return policies, and service agreements—once. Smart Disputes will then automatically select and include the appropriate documents in the evidence packet based on the dispute’s reason code, network requirements, and the cardholder’s claims.

Stripe has also published a public roadmap with detailed entries through the first quarter of 2027, including upcoming Radar features and improvements.