AI and the Future of Your Store

Part 2: AI Jailbreaks, De-Risking Your AI Tech Stack, and Marketing to AI

The AI Bots Break Loose.

AI agents at OpenAI, Anthropic, and Meta broke out of their test environments and hit the open internet. Anthropic even disclosed that its agents gained unauthorized access to the real production systems of three independent organizations. The AI didn’t use some sophisticated, Hollywood-style hacking technique. It simply exploited weak passwords and unauthenticated endpoints, obtained application and infrastructure credentials, and accessed a database containing several hundred rows of real production data.

Now, imagine that happening inside your dealership.

Scary.

Most dealerships don’t have Anthropic’s security resources. Many are operating with old or shared passwords, loosely connected systems, and little formal oversight across IT, cybersecurity, compliance, and data governance. Nonetheless, they are introducing AI agents capable of navigating those systems, taking actions, and accessing customer information at speeds no employee could match.

AI doesn’t need to become an evil super-intelligence to create a catastrophic breach. It simply needs to access a weakness your organization never fixed. When basic security hygiene is lacking, a single exposed endpoint or compromised credential could grant an AI agent access to hundreds or thousands of customer records, resulting in regulatory fines, compliance violations, and reputational damage before anyone inside the dealership even realizes what’s happening.

That is the danger of adopting AI without governance. AI doesn’t merely introduce new risks. It finds, accelerates, and magnifies vulnerabilities already hidden within your technology stack.

This is Part 2 of our series on AI and the future of your store. Last week, we covered AI, people, and culture. This week’s newsletter is about the technology: how you select it, govern it, and keep it from damaging your profitability, reputation, and legal standing.

AI to Green-Light (& What to Avoid)

Dealers should lean into AI with the understanding that every application carries a different risk profile. It’s apparent from this string of recent AI incidents that the technology always carries some risk. Some technologies offer high upside with relatively manageable risk, though. Others carry enormous risk without all that much upside. The following applications and corresponding tools are generally low-risk, high-reward. They’re relatively easy to audit and do not independently alter the economics of a transaction, which is important when it comes to keeping you out of potential trouble with the FTC.

#1 Document intake and workflow automation

Anything that’s related to automating document intake tends to be relatively safe.

The vendors I recommend include DealCheck AI by ComplyAuto, Dealertrack by Cox Automotive, and, for document automation specifically, you won’t go wrong with Tekion.

#2 Voice Scheduling

Automated, 24/7 AI voice assistants and receptionists handle inbound phone calls and book appointments directly onto business calendars. These tools risk irritating customers from time to time, but they provide unprecedented availability for clients who need to do something as simple as schedule an oil change.

The vendor I’d go with for this is Stellar.

#3 Operational AI

AI-enabled reporting and insight tools that connect to your DMS and bring financial, operational, and performance data together in one place can be a complete game-changer for your business. Especially when the tool converts that data into insights, benchmarks, and action plans that coach your people in real-time.

There’s only one tool that does this: SmartDealer.

AI to Avoid, at Least for Now

These AI tools require careful consideration. I’m not saying you can NEVER use them. I would recommend proceeding with caution, though.

#1 Dynamic pricing based on personal behavior

Adjusting a vehicle’s price based on supply/demand, market competition, condition, or age isn’t a problem. That’s inventory pricing. When a system decides that this particular customer may be willing to pay more based on what the dealership knows about them, you could have a problem. The FTC did not say all personalized pricing is illegal, but last month it said businesses may violate federal law when consumers reasonably expect a price to be broadly available, but the business quietly personalizes that price without clearly explaining that it has done so, why it has done so, and what data it used.

#2 Bots that negotiate price

A bot that follows up with a customer is one thing. A bot that independently negotiates a price, payment, trade value, or financing terms is something else entirely. Without hard guardrails and complete records, it can create promises and inconsistencies your employees are left to unwind. Given the FTC’s focus on pricing, dealers must understand not only what price they advertise, but whether their technology changes the price, offer, discount, or terms presented to an individual… and why. That’s hard to do when you have a bot negotiating on your behalf.

#3 Any black-box recommendations

If a system recommends a particular price, product, offer, or action, but neither the dealer nor the vendor can explain why, that is a problem. “Because the AI said so” is not an acceptable answer to a customer, an attorney, or a regulator.

Before approving any AI that touches pricing or payments, I would ask:

• Exactly which data fields does the AI use?
• Can we control which attributes influence the outcome?
• Does it recommend changes or make them automatically?
• Can we audit every customer-facing output?
• Can we immediately disable any high-risk functions of the tool?

If the vendor you are evaluating cannot answer clearly, do not assume the technology is just too sophisticated to explain. Assume you do not know enough to accept the risk.

Importantly, you cannot exclusively review technology when you buy it and assume it will behave the same way forever. AI systems change. Vendors change. Your data changes. Your workflows change. Governance has to change with them.

Most dealerships will bolt on AI.

Very few will architect it.

The goal is not to have the most AI.

The goal is to build the most intelligent, transparent, and accountable dealership.

That is how you turn AI from another source of chaos into a real competitive advantage.

Start Marketing to AI

As your dealership evaluates which AI tools to leverage and avoid, it’s important to recognize that your customers are also adopting AI at breakneck speeds. Customers are already using AI to research vehicles, compare ownership costs, evaluate dealerships, understand financing, and decide where to buy. That means you have to market to AI if you want to reach your customers.

When someone asks an AI system which SUV is best for their family, how much they should pay, or which nearby dealership has the strongest reputation, will it know you exist?

Will it understand what differentiates you?

Will it have enough credible information to recommend you?

Historically, dealers have invested heavily in paid marketing while publishing very little useful content of their own. That balance needs to change. Fast. Dealers need to publish useful, authoritative information about their inventory, brands, ownership costs, service questions, financing, local market, people, and processes.

AI is now in a marketing arms race. The stores that build that authority early will create efficiencies that become increasingly difficult to match. The stores that ignore it may eventually discover they cannot spend enough on traditional advertising to close the gap.

6-Step AI Playbook for Dealers

Step 1: AI Inventory

List every tool that claims to use AI, generates customer-facing communication, recommends an action, or makes an automated decision. Do not limit the list to products with “AI” in the name. Many tools already use algorithms or automated decision-making in the background without calling attention to it.

For each tool, determine:

• What problem is it solving?
• What business outcome will it improve? How will you know if it’s working?
• What data can it access?
• What decisions can it influence?
• What can it change without approval?
• Who monitors its outputs?
• Who can shut it down?

Step 2: Audit Your Data Layer

If the underlying data you have is fragmented, dirty, or inconsistent, it renders the AI tools essentially useless. Figure out what information your tools are accessing and audit those data fields relentlessly. Garbage in, garbage out.

If a tool touch prices, discounts, payments, financing, incentives, or customer eligibility, you need to be even more vigilant. The more consequential the decision, the more scrutiny the data layer deserves.

Step 3: Map Integration(s)

Which other systems does it connect to?

Step 4: Lock permissions down/reduce access

Step 5: Vet Future Vendors

Step 6: Build Continuous Governance

If the FTC knocks on your door, you cannot point at an AI tool and say, “he did it.” It’s imperative that someone at your store takes accountability for every decision being made by AI. That’s the only way to keep your dealership out of trouble. Assign an internal owner. Set an audit cadence. Document changes. Review actual outputs. Create an escalation process. Make sure someone inside your organization has both the authority and the knowledge to intervene.

Not a policy PDF collecting dust. A real system of ownership, oversight, and accountability.

Short answer: [one or two sentences that answer the headline outright, before any set-up].

Key takeaways

1. [First takeaway, stated as a finding, not a topic]
2. [Second takeaway]
3. [Third takeaway]
4. [Fourth takeaway]
5. [Fifth takeaway]

By the numbers

The standard to hold

[What has to be true before anything else works. Name the specific platforms, OEM programs, industry bodies or published research where they are not confidential.]

How the best operators run it

[The steps in order. Who owns each one, and how often it happens.]

Comparing the two common approaches

How to choose

This fits you if: [ ]

This does not fit you if: [ ]

Five questions to ask before you commit

1. [ ]
2. [ ]
3. [ ]
4. [ ]
5. [ ]

Red flags: [ ]

The mistakes that cost the most

[Two or three specific, expensive mistakes and what they cost.]

Your next three moves

1. [Action, with an owner and a date]
2. [Action]
3. [Action]

About this guidance

Written by David Spisak from nearly 50 years in retail automotive. Last reviewed [month year].

Built on DGS Labs methodology · © 2026 DGS Labs. All rights reserved.