New York, NY – November 11, 2026 – The burgeoning era of artificial intelligence in travel booking has ushered in a new era of complex questions, none perhaps more pressing than the issue of liability when AI agents err. At the Skift Commerce Summit held in New York City on Wednesday, Abhinav Anand, Citi’s Head of Value Cards, Lending, and Commerce for U.S. Consumer Cards, articulated a clear stance on this burgeoning concern: the company behind the AI agent should bear the financial responsibility for its mistakes, rather than the unsuspecting hotel, airline, or card issuer.

“Who takes the agent liability?” Anand posited, framing it as the paramount question currently being debated by a broad spectrum of industry players, including banks, payment networks, merchants, and the very AI companies developing these sophisticated booking tools. His pronouncement during his address at the prestigious Skift Commerce Summit signals a pivotal moment in the ongoing dialogue about the integration of AI into the travel ecosystem and the establishment of clear lines of accountability.

In Citi’s perspective, as articulated by Anand, the onus of responsibility for AI-induced booking errors should unequivocally rest “to the agent themselves, the company that is actually coming with the agent.” This perspective challenges the traditional allocation of risk, where errors often cascade through the transaction chain, potentially impacting various stakeholders. Anand elaborated that a potential framework for enforcing this liability could involve requiring AI agents seeking to transact on major payment networks like Visa or Mastercard to implicitly or explicitly accept a degree of accountability for their booking outcomes. This would necessitate a fundamental shift in how AI booking platforms are integrated into existing financial and travel infrastructure.

The implications of this statement are far-reaching. As AI-powered travel booking agents become increasingly sophisticated and integrated into consumer and corporate workflows, their capacity to make errors – whether due to flawed algorithms, incomplete data, or unforeseen system glitches – grows. These errors can manifest in various costly ways: booking the wrong dates or destinations, selecting incorrect flight classes or room types, failing to apply applicable discounts or loyalty program benefits, or even inadvertently creating duplicate bookings. Historically, when such errors occurred through human agents, the resolution process often involved negotiation and chargebacks, with varying degrees of success for the consumer and the merchant. However, the scale and speed at which AI agents operate introduce new complexities.

Citi’s proposed solution shifts the burden from the downstream service providers – the hotels and airlines that fulfill the booking – and the financial intermediaries, to the entity that developed and deployed the AI. This aligns with a principle of holding the originator of the service accountable for its performance and its potential failures. Anand’s remarks suggest that Citi, as a major card issuer and financial institution, is actively considering how to structure its partnerships and transaction protocols to reflect this liability model.

To further illustrate the magnitude of this issue, consider the current landscape of AI in travel. Companies are investing billions in developing AI agents capable of understanding natural language commands, personalizing recommendations, and automating complex itineraries. These agents are being integrated into everything from corporate travel management platforms and consumer-facing booking apps to virtual assistants and smart home devices. The promise is a seamless, efficient, and highly personalized travel planning experience. However, the potential for AI to misinterpret a user’s intent, access outdated information, or encounter a technical malfunction is a tangible risk.

For instance, an AI agent tasked with booking a business trip might misunderstand a request for “a quiet hotel near the convention center” and instead book a hotel in a noisy entertainment district, leading to a disgruntled traveler and potential lost productivity. Or, an AI might fail to recognize a traveler’s elite status with a particular airline, resulting in missed upgrade opportunities or incorrect fare classes being booked. In such scenarios, the hotel might face a guest dispute, the airline might have to re-accommodate a passenger, and the card issuer might be asked to initiate a chargeback. Under the traditional model, these parties would bear the brunt of the resolution process, often absorbing costs and spending valuable time on customer service.

Citi’s proposed framework suggests a different path. If an AI agent booked the incorrect hotel, the company that developed and deployed that AI agent would be liable for the costs associated with rectifying the mistake. This could include the difference in price for the correct hotel, any cancellation fees incurred, and potentially compensation for the traveler’s inconvenience. This approach incentivizes AI developers to invest heavily in the accuracy, reliability, and robustness of their agents. It also necessitates the development of clear service-level agreements (SLAs) between AI providers and the platforms through which they operate, whether those are payment networks or direct integration partners.

The mention of Visa and Mastercard is particularly significant. These global payment networks are the gatekeepers for a vast majority of transactions. Their willingness to implement policies that require AI agents to accept liability would be a powerful catalyst for change. It could involve new contractual obligations for companies offering AI booking services, or even the development of specific AI liability insurance products that would be mandated for participation in their networks. Such requirements would undoubtedly lead to more rigorous testing and validation of AI booking algorithms before they are deployed to consumers and businesses.

The current lack of a clear liability framework presents a significant hurdle to the widespread and confident adoption of AI in travel. Businesses are hesitant to fully embrace AI-driven booking solutions if they fear being held responsible for unforeseen AI errors. Travelers, too, may be wary of relying on AI for critical bookings if they are unsure who to turn to in the event of a problem. Citi’s proactive stance aims to address this uncertainty and foster greater trust in AI-powered travel commerce.

Beyond the immediate financial implications, the question of agent liability also touches upon broader issues of consumer protection, regulatory oversight, and the future of the travel industry. As AI becomes more autonomous, the distinction between a tool and an agent blurs. Establishing accountability for AI actions is crucial for maintaining consumer confidence and ensuring a fair and functional marketplace.

Anand’s address highlighted that this is not a hypothetical debate for Citi. The bank, as a major player in both consumer and commercial lending and payments, is directly exposed to the financial ramifications of booking errors. By taking a firm stance, Citi is signaling its intention to shape the future of AI in commerce and to protect its interests and those of its customers.

The path forward likely involves a multi-faceted approach. AI companies will need to invest in advanced error detection and correction mechanisms, robust data governance, and transparent communication protocols. Payment networks will need to adapt their infrastructure and contractual frameworks. And regulatory bodies may eventually need to step in to establish clear guidelines and consumer protection standards for AI-driven commerce.

Citi’s leadership in raising this critical question and proposing a clear direction is a welcome development for an industry on the cusp of a major technological transformation. The future of travel booking promises to be more intelligent and personalized, but it must also be secure, reliable, and accountable. By placing the responsibility squarely on the shoulders of the AI providers, Citi is advocating for a future where innovation in AI is matched by a commitment to its responsible deployment and a clear understanding of who pays when things go wrong. The industry is now watching to see how other major players will respond to this pivotal challenge. The conversation has begun, and the stakes are incredibly high.