AI, Risk, and Recovery in Auto Finance

WHAT INDUSTRY LEADERS TOOK AWAY FROM A CLOSED-DOOR ROUNDTABLE AT USED CAR WEEK 2025

In November 2025, Allied Solutions convened senior auto finance leaders at Used Car Week for a closed-door roundtable. The purpose was simple: to compare notes on what is meaningfully changing across lending, servicing, and recovery, and what remains harder to move despite new tools and increased attention.
The discussion touched on artificial intelligence, fraud, repossession strategy, compliance oversight, and portfolio performance. What stood out was the tone. There was little appetite for bold predictions or sweeping claims. Instead, participants focused on what is working today, where progress has been uneven, and how institutions are adjusting in practical ways.
The insights below reflect those conversations and the perspectives of organizations operating at scale.

How Executives Are Using AI Today
The conversation around AI has shifted. Leaders no longer framed it as a looming threat or a shortcut to transformation. Instead, they spoke about fit. Where AI integrates cleanly into existing workflows. Where it introduces risk. Where it genuinely helps people make better decisions.
Most institutions described a similar approach. They are testing narrow use cases, validating outcomes, and building governance alongside deployment. Broad automation is not the goal. Learning is.
One theme came up repeatedly. Fluency matters more than speed. Knowing when to use AI, when not to, and how to remain accountable for outcomes.
Questions raised during the discussion included:

Where does AI clearly improve decision quality?
How much automation is appropriate in a regulated environment?
Which capabilities create lasting advantage rather than short-term efficiency?

Where AI Is Adding Operational Value
Rather than novelty, leaders pointed to applications that reduce friction in long-standing processes.
Recovery intelligence was a common example. Several organizations are using AI to surface risk earlier and allocate resources more effectively, including:

Identifying accounts unlikely to reach repossession
Flagging recoveries that are likely to be more complex due to impounds, title issues, or redemption risk
Prioritizing actions based on cost, probability, and timing

Vehicle condition and remarketing was another area of momentum. Participants shared examples of AI being used to:

Review condition reports and images to support consistent grading
Identify reconditioning investments that materially affect sale outcomes
Validate invoices before payment to reduce leakage

Auction and logistics decisions are also becoming more data-informed. Leaders described using historical performance, vehicle attributes, and localized demand data to answer a persistent operational question at scale. Once a vehicle is recovered, where should it go next?
Across these use cases, a consistent view emerged. AI functions best as a prioritization tool. Final decisions still belong with people.

Recovery Strategy: Precision Over Volume
Repossession volumes remain elevated, but leaders agreed that the broader recovery ecosystem has adjusted. Transport capacity, auction throughput, and vendor availability are no longer the primary constraints they were in recent years. Vehicles can be located and moved.
The challenge has shifted.
Improving Outcomes After Recovery Begins
The harder question is how to improve results once recovery is underway. Several participants referenced the difficulty of consistently moving beyond the industry’s typical net recovery range.
AI-driven insights are increasingly viewed as one lever to address this, particularly when applied to:

Timing decisions that affect downstream cost
Avoidable expenses that erode net proceeds
Condition management and sale execution

Leaders also pointed to total loss insurance and related recovery adjacencies as areas that deserve more attention. These functions are not always treated as part of the recovery strategy, but they have a direct impact on margins.
The takeaway was clear. Recovery performance depends less on effort and more on coordination.

Fraud as a Lifecycle Risk
Fraud was described as persistent and structural. Participants emphasized that it now spans the full loan lifecycle, from origination through servicing and recovery. Identifying signals earlier and acting faster has become a shared responsibility across teams.
Pressure points discussed during the roundtable included:
Dealer and transactional fraud, such as:

Straw purchases and identity misuse discovered after funding
Dealer buy-back negotiations once fraud is confirmed
The operational and legal complexity of unwinding bad paper

Impound, lien, and title-related fraud, including:

Organized impound schemes and lien manipulation
Notices mailed blank but still signed for
Elevated lien loss exposure in states such as Florida, Texas, and New York, particularly when third parties are involved

Payment manipulation, where large payments are posted and reversed in quick succession, often leaving lenders exposed while disputes are resolved.
Technology Surfaces Risk. People Decide.
While analytics and AI can surface anomalies more quickly, leaders were consistent on one point. Judgment at the frontline remains essential. Several institutions shared examples of recognizing or rewarding employees who identify fraud early.
In practice, fraud mitigation remains a hybrid discipline. Technology accelerates detection. People determine the response.

Servicing, Compliance, and the Borrower Experience
Participants expressed growing interest in AI-enabled servicing tools that reduce friction for borrowers and support staff.
Examples discussed included:

AI-supported call centers and mobile interfaces
Reduced reliance on traditional IVR systems
Clearer, more intuitive responses to common borrower questions

AI-driven welcome calls were cited as a particularly effective use case. Westlake’s pilot was referenced as an example of how early engagement can be scaled without sacrificing consistency.
From Basic Support to More Nuanced Interaction
Most current deployments focus on straightforward inquiries. That said, leaders see a path toward more complex borrower interactions as models mature and regulatory confidence increases.
Compliance as an Operational Application
AI is also being applied to oversight and governance, including:

Monitoring call tone and language
Flagging potential discrimination or inappropriate conduct
Supporting quality assurance across large servicing operations

This positions AI not only as an efficiency tool, but as a way to strengthen risk management.

Portfolio Performance: Pressure Now, Gradual Improvement Ahead
Leaders largely agreed that portfolio performance is unlikely to rebound quickly. Expectations instead point to a gradual improvement.
Near-term pressure remains across delinquencies, repossessions, and losses. Several participants anticipate a plateau forming through 2026, with more meaningful stabilization emerging later in the year. Interest rate relief was viewed as helpful, but not immediate.
Structural challenges persist. Bankruptcy filings remain elevated, and courts often favor consumers. Loans originated between 2021 and 2023 continue to underperform, reflecting credit profiles shaped by unusual conditions.
The conclusion was consistent across organizations. Traditional indicators alone are no longer sufficient. Deeper behavioral and performance analytics are required to understand borrower risk.

What This Signals for Auto Finance Leaders
Several realities were reinforced during the roundtable.

AI adoption is progressing, but value depends on discipline and specificity. Choosing the right problems to solve matters more than broad automation.
Recovery performance is increasingly about precision. Volume alone does not drive outcomes.
Fraud prevention requires visibility across the full lifecycle and teams that are empowered to act.
Servicing, compliance, and borrower experience are no longer separate conversations. Decisions in one area increasingly shape the others.
Portfolio normalization will take time, even as macro conditions begin to improve.

The next phase of auto finance is unlikely to reward speed for its own sake. It will favor organizations that apply judgment carefully, supported by intelligence where it adds measurable value.