AI Companion Payment Processing: The Complete Guide for Platform Founders

The AI companion market is crossing $1 billion in annual revenue. The payments infrastructure available to it was not built for it.

Founders are losing processor accounts with no explanation, burning runway on failed applications, and making costly structural decisions based on generic high-risk acquiring advice that doesn’t apply to their category. Meanwhile, the acquirers best positioned to serve this market are issuing blanket declines because their underwriting frameworks were designed for a world without AI-generated companions.

This guide addresses that gap entirely. What you will find here: the real reason AI companion platforms are being declined, a tier-by-tier analysis of the acquirer spectrum, original survey findings from 12 acquirers, a proprietary readiness framework for evaluating your platform before you apply, and the MCC classification landscape with a clear-eyed view of where it is heading.

Why AI Companion Platforms Get Declined, and It’s Not What You Think

If you have been through a processor decline or termination, you have probably been told very little about why it happened. A brief email, a reference to terms of service, perhaps a mention of “business type.” What you almost certainly have not been told is the actual mechanism that produced the outcome.

The Reputational Risk Misclassification Problem

Payment processors operate within a tiered risk framework. At the top sits reputational risk: businesses that may be technically legal but whose association could damage the processor’s relationships with card brands, correspondent banking partners, or regulated financial institutions.

When an underwriter receives an application from an AI companion platform and searches the product, their pattern-matching, calibrated against years of underwriting adult content merchants, fires immediately. The application gets routed to the wrong team, evaluated against the wrong criteria, and declined for failing standards it was never actually subject to. The platform has not failed an appropriate test. It has been graded on the wrong rubric.

What Acquirers Actually Fear (Hint: It’s Not Chargebacks)

Ask a founder why processors reject their platform and the answer is almost always: chargebacks. Ask an underwriter the same question about AI companion platforms and the answer is almost always: reputational exposure, content liability, and regulatory uncertainty.

• Reputational exposure: the acquirer is worried that processing payments for your platform could attract negative press coverage or card brand scrutiny, regardless of your actual chargeback rate.
• Content liability: the acquirer cannot determine whether your platform will remain within the content boundaries it currently operates, or whether user-generated interaction patterns will produce outcomes that create legal exposure.
• Regulatory uncertainty: AI companion is a fast-moving category with no established compliance framework, and underwriters cannot accurately price that risk.

None of these concerns disappears because your chargeback rate is 0.4%.

The Conflation with Adult Content and Why It’s Killing Approvals

Acquirers use MCC 5967, the code most commonly associated with explicit adult content, as a trigger for elevated scrutiny. The documentation requirements include age verification protocols, content moderation infrastructure, and performer consent records designed specifically for explicit content platforms.

An AI companion platform that has none of this documentation, because it produces none of the content that documentation was designed to govern, fails that checklist and gets declined. Not because it is deficient. Because it is being evaluated against criteria for a different business. This conflation is the single most expensive problem in AI companion payment processing.

How Acquirers Evaluate AI Companion Platforms: The Underwriting Reality

The Six Underwriting Signals Acquirers Look For

Based on the acquirer survey conducted for this article, the underwriting evaluation for AI companion platforms consistently focuses on six signals, in roughly this order of weight:

1. Content boundary clarity: Does the platform have clear, technically enforced limits on the type of content produced by or available through the AI? Are those limits documented as product architecture decisions rather than just terms of service aspirations?
2. Business legitimacy indicators: How long has the company been operating? Is it incorporated in a recognised jurisdiction? Does it have verifiable team members, app store listings, press coverage, or investor backing?
3. Compliance infrastructure: Does the platform have a credible age verification mechanism? Are its terms of service clear, fair, and written in plain language? Is subscription cancellation genuinely easy to execute?
4. Revenue model transparency: Are charges clearly communicated at the point of purchase? Does your subscription billing model meet the FTC’s updated Negative Option Rule requirements? Are virtual currency prices displayed in fiat equivalent?
5. Chargeback and dispute readiness: Not chargeback rate, readiness. Is there tooling in place to monitor, manage, and respond to disputes? Are transaction records maintained in a format suitable for representment?
6. Processing history: Prior terminations, especially those entered into the TMF/MATCH list, are significant negative signals.

What “Grey Area” Means in Acquiring Terms

The phrase “grey area merchant” has a specific meaning in acquiring. It refers to a merchant whose business is legal and whose product does not violate card brand rules outright, but whose category lacks established underwriting criteria and whose risk profile cannot be accurately modelled from existing data. Grey area merchants are not declined because they are bad actors. They are declined because acquirers cannot price the risk with confidence.

AI companion is currently a grey area category in acquiring. There is no established chargeback benchmark for the category. There is no MCC specifically assigned to it. The framework is still being written, which creates both risk and opportunity for founders willing to engage constructively.

The Due Diligence Walk-Through: What Happens After You Apply

When an AI companion platform submits a merchant application, the process typically unfolds across three stages:

Initial screening (24 to 72 hours). The underwriting team reviews your application, checks principals against the MATCH list and OFAC databases, visits your website and app store listing, and makes an initial category classification. This is where the misclassification problem most commonly occurs.

Substantive underwriting review (3 to 10 business days). The underwriter reviews supporting documentation, may perform a deeper product review, and may submit to the acquiring bank’s compliance team for secondary review on reputational grounds.

Commercial terms negotiation. If both reviews are satisfied, processing rates, reserve structure, and volume limits are established. For AI companion platforms, reserve terms are frequently the most consequential negotiation. Informed founders do significantly better than uninformed ones.

MCC Classification for AI Companion Platforms: Where You Actually Sit

The Merchant Category Code assigned to your account affects your interchange rates, the card brand rules that govern your transactions, your chargeback dispute resolution pathway, and whether your transactions get routed to adult content monitoring programs. This matters enormously and is almost entirely misunderstood by founders.

The MCC Codes in Play

Ordered best-case to worst-case for a non-explicit AI companion platform.

Why MCC Classification Is Not Your Choice to Make

MCC classification is assigned by the acquiring bank, not selected by the merchant. You can, and should, make the case for appropriate classification in your application, but the acquirer makes the determination. The practical implication: your application materials should proactively address classification, making the explicit case for why your platform is appropriately classified as software or SaaS, with specific evidence (content policies, technical architecture, compliance documentation) that supports that classification.

The Classification Battle Coming in the Next 24 Months

Both Visa and Mastercard are monitoring the AI companion category. The market is too large and too fast-growing to be ignored, and the current state, in which different acquirers assign different MCCs to functionally identical platforms, creates inconsistency that card brands typically resolve by issuing explicit guidance.

That guidance, when it comes, could create a dedicated MCC for AI companion platforms, issue explicit classification criteria routing platforms to 7372 or 5967 based on content, or extend adult content registration requirements to platforms that include romantic or intimate AI interactions regardless of explicit content. Founders building for the long term need platforms that are defensible under any of these scenarios.

The Acquirer Spectrum: Who Will Actually Process Your Payments

Four tiers of processor will look at an AI companion platform, and the right target depends on how much of the readiness work you have already done. Rates and reserves climb as you move down the tiers.

Tier 2 is the most commercially attractive realistic target for a compliant platform. Reserves held 90 to 180 days.

Tier 1: Payment Facilitators, The Honest Assessment

Stripe, PayPal, and Square operate as payment facilitators. They aggregate merchants under their own master merchant accounts and onboard businesses at scale with automated approval systems and minimal human underwriting. Their terms of service explicitly prohibit categories they have designated as unacceptable, and for most AI companion platforms, particularly those with romantic, companion, or intimate interaction features, that prohibition applies.

The honest assessment: for most AI companion platforms, Stripe, PayPal, and Square are not currently viable primary processors. Accounts that pass initial automated screening are at significant risk of termination following manual review or a customer complaint trigger. This is not permanent, payment facilitators follow market size, but founders should not build their revenue infrastructure on a foundation that can disappear with 24 hours’ notice.

Tier 2: Domestic ISOs and Acquirers, What Conditions Unlock Access

Domestic ISOs and their sponsoring acquiring banks represent the most commercially attractive realistic target for most AI companion platforms. These organisations underwrite merchants with more human involvement and greater category nuance than payment facilitators. Some have already developed working frameworks for AI companion platforms.

Access to Tier 2 typically requires meeting several conditions simultaneously: no explicit content with technically enforced boundaries; compliance infrastructure meeting the baseline described in the URF framework below; a clear subscription or monetisation model without dark pattern risk; and business legitimacy signals that distinguish a professional operation from a fly-by-night platform. Processing rates at Tier 2 typically range from 2.9% to 4.5%, with reserve requirements of 0% to 10% rolling.

Tier 3: Specialist High-Risk Acquirers, What to Expect on Rates and Reserves

Specialist high-risk acquirers exist specifically to process merchants that domestic ISOs decline. They have category expertise in adult content, firearms, nutraceuticals, gaming, and, increasingly, AI content platforms. For AI companion platforms that cannot yet meet Tier 2 requirements, Tier 3 is the appropriate home, not a failure, but an appropriate current position.

What to expect at Tier 3: processing rates typically ranging from 4.5% to 7.5%. Rolling reserves of 10% to 25%, held for 90 to 180 days, are standard. Volume caps in the early relationship are common. Tier 3 is also where some of the worst commercial terms in the industry exist. Rate structures with ambiguous fee stacking, rolling reserves with vague release conditions, and punitive early termination clauses are more common here. Due diligence on the acquirer is essential before signing.

Tier 4: Offshore Acquiring, When It’s Legitimate and When It’s a Trap

Offshore acquiring refers to processing through acquiring banks domiciled outside the merchant’s home jurisdiction, typically in jurisdictions with more permissive financial regulatory environments.

Alternative Rails: ACH, Crypto, and Emerging AI Payment Infrastructure

ACH processing sidesteps card network rules entirely and can be a useful supplement for high-value transactions from verified US bank accounts. It is not appropriate as a primary payment method for international customers or low-friction consumer purchases, but it provides meaningful processing diversification for subscription renewals.

Crypto payment rails offer card-network-independent processing with global reach and no MCC classification problem. The practical limitations are significant: consumer adoption remains niche, volatility creates accounting complexity, and the friction of crypto checkout meaningfully reduces conversion rates for mainstream consumer audiences.

The TPE Underwriting Readiness Framework for AI Companion Platforms

The TPE Underwriting Readiness Framework (URF) was built specifically for AI companion platforms to address a gap in available tools: there was no structured way for founders to evaluate their own processing readiness before spending time and reputational capital on applications they would fail.

The URF evaluates a platform across five domains, each scored on a 1 to 4 scale, producing a composite score between 5 and 20 that maps to a recommended processing tier and identifies specific remediation priorities.

The Five Readiness Domains

Domain 1: Content Boundary Architecture (CBA)

This domain evaluates whether content limits are technically enforced or merely stated in policy. Score 1: no explicit content restrictions and no moderation infrastructure. Score 2: terms of service restrictions without technical enforcement. Score 3: technically enforced content filters with documented content policy. Score 4: explicit content entirely absent from platform architecture, with enforced filters, documented content policy, and audit logs demonstrating policy enforcement over time. Acquirers look here first.

Domain 2: Compliance Infrastructure (CI)

This domain covers age verification strength, KYB and AML policy documentation, terms of service quality, and data privacy posture. A score of 4 indicates robust age verification meeting applicable state law requirements, clear and legally reviewed terms of service with plain-language cancellation and refund processes, and documented GDPR and CCPA compliance. Platforms operating in states with enacted age verification requirements, including Texas and Arkansas, need to demonstrate compliance with those specific statutes.

Domain 3: Revenue Model Transparency (RMT)

This domain evaluates how clearly charges are communicated at point of purchase. A score of 4 indicates full compliance with the FTC’s Negative Option Rule, fiat equivalent pricing displayed at all virtual currency purchase points, clear subscription terms with one-click cancellation, and billing descriptors that accurately identify the merchant. Virtual currency economies receive particular scrutiny because their pricing is often opaque by design.

Domain 4: Chargeback and Dispute Readiness (CDR)

This domain evaluates operational preparedness for dispute management, not historical chargeback rate. A score of 4 indicates a dedicated dispute management tool integrated with the processor, documented representment processes for the three most common dispute reason codes in the AI companion category (subscription confusion, unrecognised charge, and dissatisfaction-based disputes), and proactive chargeback rate monitoring with defined internal response thresholds.

Domain 5: Business Legitimacy Signals (BLS)

This domain evaluates the indicators that distinguish a professional operation from a high-risk opportunistic platform. A score of 4 indicates registered company formation in a credible jurisdiction, verifiable principals with professional online presence, 12 or more months of operating history, app store listings demonstrating user adoption, press coverage from credible sources, and investor backing or funding documentation if applicable. The absence of these signals is a fast path to decline.

How to Score Your Platform Before You Apply

TPE Underwriting Readiness Framework. Address any single domain scoring below 3 before applying.

The Application Package: What to Include and How to Frame It

A complete merchant application for an AI companion platform should include: company documentation (certificate of incorporation, proof of registered business address, principal ownership structure, government-issued ID for all principals); a platform description that explicitly addresses the adult content conflation concern with evidence; content policy documentation covering the actual written policy as implemented in the product; compliance documentation including age verification methodology, terms of service, and cancellation process; revenue model disclosure; and processing history.

What 12 Acquirers Actually Said: Original Survey Findings

The survey conducted for this article reached 12 acquiring institutions across the high-risk acquiring spectrum: four domestic ISOs with digital goods specialisation, five specialist high-risk processors with adult content category experience, two offshore acquirers with US market exposure, and one emerging AI-specific payment infrastructure provider. All participants responded on the condition of anonymity. The survey was conducted over a six-week period in mid-2025.

Key Finding 1: The Content Clarity Threshold

Nine of the 12 acquirers surveyed identified content boundary clarity as the single most important factor differentiating declined applications from conditionally approved ones. Not chargeback rate. Not processing history. Content clarity.

If I open the app and the companion starts saying things that could be romantic, and I can’t find a policy that tells me where the line is, I have to decline.

Domestic ISO underwriter, survey respondent

Key Finding 2: Reserve Expectations by Platform Type

Reserve requirements varied significantly by platform type. The reason, according to three separate acquirers, is dispute complexity: subscription disputes have a well-understood anatomy and established representment pathway, while virtual currency disputes are harder to resolve because the relationship between fiat payment and consumed value is opaque.

Reserve release timelines ranged 90 to 180 days, contingent on maintaining chargeback rates below 0.65%. Original TPE acquirer survey, mid-2025.

Key Finding 3: The Features That Move You From Decline to Conditional Approval

Five platform features or documentation elements most consistently moved an application from outright decline to conditional approval:

• A documented and technically enforced age verification mechanism that produces a verification record usable in dispute resolution
• The ability to produce a timestamped content log of AI interactions for a given user, demonstrating content boundary compliance
• A billing descriptor management system that produces recognisable merchant names on cardholder statements
• A clear, one-click cancellation mechanism with a confirmation email
• A demonstrated chargeback monitoring process, not just a low rate, but evidence that the rate is being actively managed

“It’s not about whether the rate is low today. It’s about whether the merchant will know when it starts going up and do something about it before we have to.”

Specialist acquirer, survey respondent

Platform Design Decisions That Affect Payment Processing

Payment processing is not only a business development problem. It is a product problem. Several design decisions made at the platform level have direct consequences for processing approvability that founders often discover only after a termination.

Age Verification Architecture

A basic age gate, a checkbox or a birth date entry field, is insufficient. It does not verify age; it requests a self-declaration. Acquirers and regulators both understand this distinction. Meaningful age verification involves identity document verification, third-party age verification services, or credit card-based age inference combined with additional verification.

Subscription vs. Token vs. Tipping: How Revenue Model Affects Underwriting

Subscription models with clear pricing, disclosed renewal terms, and easy cancellation are the most acquirer-friendly revenue structure in this category. They produce predictable revenue, well-understood dispute patterns, and a clean compliance interface with the FTC’s Negative Option Rule.

Token and virtual currency models create underwriting complexity because the relationship between purchase price and value received is non-linear and often opaque to consumers. Dispute rates in token economy platforms are structurally higher, acquirer appetite is lower, and reserve expectations are higher. Virtual gifting and tipping features add a social payment dynamic. When a user tips an AI companion, they are making a discretionary payment with no guaranteed return, creating buyer’s remorse risk that is distinct from subscription confusion.

Terms of Service and Refund Policy Requirements

Terms of service are reviewed by underwriters as a governance document, not just a legal one. An underwriter reading your terms is asking: does this company understand what it is doing? Has it thought carefully about consumer protection? A strong terms of service document clearly describes the nature of the AI interaction (including explicit disclosure that the companion is an AI, not a human), subscription terms and renewal cadence, refund eligibility and process, content boundaries and moderation policy, and account termination criteria.

A refund policy that provides genuine recourse for consumers, not a blanket “all sales final” stance, is consistently preferred by acquirers. The practical reasoning is simple: a consumer with a refund pathway is less likely to file a chargeback.

The Contrarian View: The Real Risk Isn’t Your Chargebacks

The conventional wisdom in AI companion founder communities attributes processor problems primarily to chargeback rates. Build a good product, reduce subscriber churn, minimise confusion at the billing moment, and your processing problems will resolve.

This is wrong. And believing it causes founders to invest in the wrong solutions.

Chargeback rates at well-run AI companion platforms are frequently comparable to or better than those at legitimate SaaS companies. The acquirer survey data collected for this article supports it directly: content boundary clarity was identified as the primary approval driver nine times out of twelve. Chargeback history was cited as a primary driver twice, and in both cases the relevant concern was the absence of chargeback monitoring infrastructure, not the rate itself.

Platforms getting declined are not, in most cases, being declined because of their actual chargeback performance. They are being declined because underwriters cannot accurately categorise them. A 0.3% chargeback rate attached to a misclassified application is still a decline.

The uncomfortable corollary for acquirers: the underwriting frameworks most processing institutions use for digital content were built for a world without AI-generated companions, and they are producing systematically incorrect outcomes for a legitimate and growing merchant category. Acquirers who invest in building category-specific frameworks for AI companion platforms before their competitors do will capture a high-value, high-loyalty merchant segment while the rest of the industry issues blanket declines.

Keeping Your Processor: Chargeback Management for AI Companion Platforms

The Chargeback Thresholds That Matter

Visa’s Visa Acquirer Monitoring Program (VAMP) sets an Early Warning threshold at 0.65% chargeback-to-transaction and a merchant ratio threshold of 0.9%. Mastercard’s Business Risk Assessment and Monitoring (BRAM) program designates merchants as High Chargeback Merchants at 1.0% and Excessive Chargeback Merchants at 1.5%. Exceeding these thresholds can result in fines passed through from your acquirer, mandatory chargeback reduction plans, increased reserve requirements, and ultimately account termination. Your acquirer’s specific thresholds, which may be stricter than the card brand minimums, should be explicitly understood before you sign any processing agreement.

Why Subscription AI Platforms Face Specific Chargeback Patterns

AI companion subscription platforms generate chargebacks through mechanisms partially distinct from other subscription businesses. The most common dispute patterns are subscription confusion (the cardholder doesn’t recognise the billing descriptor or has forgotten they subscribed), dissatisfaction-based disputes, and friendly fraud. An AI companion platform that sends a reminder email three days before a subscription renewal and provides a one-click cancellation link will experience materially fewer subscription confusion disputes than one that does not.

Dispute Management Tools and Representment Strategies

Chargeback management tools, including Verifi (owned by Visa) and Ethoca (owned by Mastercard), provide pre-dispute notification that allows merchants to issue refunds before a dispute is formally filed, removing it from the chargeback count. Effective representment for AI companion platforms requires maintaining transaction records, session activity logs, consent records, and communication history for a minimum of 24 months.

The Regulatory and Card Brand Trajectory: Building for What’s Coming

1. 12 monthsHigh confidence

   Niche acquirer emergence and category ban expansion

   Specialist acquirers bring AI-specific underwriting frameworks to market; founders engaging now get first access. Payment facilitators that have not yet issued explicit AI companion bans will do so as compliance teams catch up.

2. 24 months

   MCC classification battles at the card brand level

   Visa and Mastercard issue formal MCC guidance; its direction (software classification vs adult rules extension) is the most consequential unknown. State age verification expands to 15+ states, becoming a near-universal underwriting requirement.

3. 5 years

   Dedicated AI payment infrastructure outside card networks

   Purpose-built rails for AI-native monetisation mature: embedded wallets, agent-to-agent payment protocols, and dispute resolution that does not rely on card brand chargeback rules.

How to Structure Your Processing Stack to Survive Each Phase

• Build compliance infrastructure to the emerging standard now, not the current minimum. Age verification, content boundary architecture, and FTC Negative Option compliance are heading toward being non-negotiable.
• Diversify your processing stack across tiers while you have the stability to do it carefully. A termination is not the right moment to discover your backup processor.
• Engage constructively with acquirers and card brand representatives building AI-specific frameworks. Founders who provide honest input influence the outcome more than founders who ignore the process and react to it.
• Maintain clean processing history through the classification uncertainty period. A strong track record at a Tier 3 acquirer over the next 24 months is the credential that opens Tier 2 when category frameworks mature.

Frequently Asked Questions

Conclusion

The AI companion market is not facing an unsolvable payments problem. It is facing a categorisation problem in an industry that builds frameworks after markets, not before them.

The platforms getting declined are not predominantly failing because of poor chargeback rates or products that genuinely violate card brand rules. They are failing because the acquiring industry does not yet have a standard framework for evaluating them, and the default response to an unfamiliar category, pattern-match to the nearest prohibited category and decline, is producing systematically wrong outcomes for legitimate merchants.

That gap is closing. Acquirers are building AI-specific frameworks. Card brands are monitoring the category. Niche payment infrastructure is emerging. The question for founders is not whether the payment landscape will develop for this category, it will, but whether their platform will be positioned to access better processing as the landscape matures.

The work that positions a platform well is the same work that builds a better product: clear content boundaries enforced technically rather than just stated in policy; transparent billing that treats users fairly; compliance infrastructure that meets emerging regulatory standards; and a business that looks, to an outside evaluator, like the serious operation it needs to be.

Next Steps: The AI Companion Merchant Processing Checklist

1. Score your platform. Run the TPE Underwriting Readiness Framework across all five domains. Address any domain scoring 2 or below before applying.
2. Assemble your application package. Company documentation, platform description with content policy disclosure, compliance documentation, revenue model disclosure, and processing history.
3. Identify your target tier. Base it on your composite URF score, from the scoring table above.
4. Prepare acquirer outreach. Use professional framing that proactively addresses the adult content conflation concern.
5. Audit billing and cancellation. Review your descriptor, subscription disclosure, and cancellation flow against FTC Negative Option Rule requirements before your application is reviewed.

Sources
1. Original TPE acquirer survey (n=12, anonymous), conducted mid-2025. Findings on file with The Payments Edge.
2. US Federal Trade Commission, “Negative Option Rule,” ftc.gov.
3. Visa, “Visa Acquirer Monitoring Program (VAMP)” overview, visa.com.
4. Mastercard, “Business Risk Assessment and Mitigation (BRAM) program,” mastercard.us.