“The Cost of Deception” is Plummeting

When discussing AI, the conversation often revolves around “what it can do.” However, what deserves the most attention now is “how much the cost of misuse has decreased.”

Fake photos for insurance fraud can be generated in minutes. Cost: nearly zero.
Fabricating 100 false complaints to take down a competing store. Cost: nearly zero.
Mass-producing malicious advertisements that erode trust. Cost: nearly zero.

In the past, fraud had a brake called “effort.” Creating fake photos required Photoshop skills. Submitting 100 false complaints needed the cooperation of 100 people. That brake has been removed by AI.

What happens when the cost for attackers approaches nearly zero? Let’s consider three case studies.

—

Case Study 1: Defrauding Insurance Companies with AI-Generated Images

According to a report by the BBC, insurance fraud using AI-generated images has increased by about 30% compared to the previous year. The method is simple: generate a photo of a traffic accident using AI and attach it to an insurance claim.

Previously, faking accident photos required a certain level of skill and time. It involved either physically damaging a car or editing photos with Photoshop. Both methods were labor-intensive and likely to leave traces.

Now, however, if you input “a car with a severely damaged front bumper” into an image generation AI, a photo can be produced in seconds. The cost is just a few yen for API usage. Even if you were to run it solely on a paid ChatGPT plan (around $20 per month), the “cost of forgery” per incident is virtually zero.

This issue is not just a problem for insurance companies. Insurance premiums are set by factoring in the costs of fraud. If fraud increases, the insurance premiums for small and medium enterprises that use insurance legitimately will rise. In other words, the burden of someone misusing AI will fall on small and medium enterprises that have done nothing wrong.

—

Case Study 2: Destroying a Competing Store with 100 False Complaints

The long-established nightclub “Heaven” in London was forced to cease operations due to a large number of AI-generated false complaints. The perpetrator was a businessman who has since been convicted.

What he did was create a large number of “non-existent complainants” using AI and flood the authorities with complaints. Names, addresses, and complaint texts — all generated by AI. The authorities lacked the resources to verify each complaint’s authenticity, ultimately leading to the decision to suspend operations.

The essence of this incident lies in the asymmetry between “the cost of attack” and “the cost of defense.”

• Cost to generate 100 false complaints: a few minutes, a few hundred yen
• Cost to recover from a suspension: legal fees, lost sales, brand damage amounting to millions to tens of millions of yen

This is not just an issue for restaurants and nightclubs. Google Maps reviews, social media reputations, and reports to authorities — all forms of “reputation” have become vulnerable to AI attacks. For local small and medium enterprises, reviews and reputations are lifelines. While large companies can rely on PR teams to manage crises, a company with just 10 employees lacks such resources.

—

Case Study 3: 8.3 Billion Malicious Ads Blocked but Still Flooding In

Google announced that it blocked approximately 8.3 billion malicious ads in 2025. 8.3 billion. It’s an astonishing number, yet ads that “slip through” continue to appear.

Why? Because AI can generate an infinite variety of ads. If one is blocked, a slightly altered version can be deployed immediately. What a human could manage with a limit of 10 patterns a day, AI can produce 1,000 patterns in an hour.

The impact on small and medium enterprises is twofold.

First, the risk of their ads being displayed alongside malicious ads. If customers perceive that “ads in this industry are suspicious,” even legitimate advertisers can suffer collateral damage.

Second, the risk of fake ads impersonating their business. Fake ads using the company name or logo can be mass-produced by AI. If customers fall victim to these fake sites, their resentment will be directed at the real company.

—

What is Happening Structurally?

The commonality among these three case studies is the structure of “plummeting attack costs” and “static (or rising) defense costs.”

As this asymmetry widens, the number of small and medium enterprises that suffer damage without doing anything will increase. This is the issue of defense costs in the AI era.

—

So, What Can Small and Medium Enterprises Do?

Simply advocating for “promoting AI education” will not change anything. Specifically, here are three actionable steps that can be taken starting today.

1. Search for “Your Company Name” Daily (Cost: 0 yen, Time: 5 minutes)

Register your company name, representative name, and main product name with Google Alerts. It’s free. Detecting fake ads, false reviews, or impersonating accounts a day earlier or later can significantly change the amount of damage incurred.

If you have extra time, check Google Maps reviews weekly and report suspicious reviews (mass postings on the same day, vague low ratings, etc.) to Google. This can also be done for free.

2. Create a “System to Preserve Evidence” (Cost: Monthly several thousand yen)

To counter false complaints and fake reviews, having evidence that “we are operating legitimately” becomes a weapon.

• For stores, save security camera footage to the cloud (monthly cost of 2,000 to 5,000 yen)
• Automatically save customer interaction emails and chat logs
• Keep transaction records with timestamps

While it may seem mundane, this becomes the material for counterattack “when something happens.” In the Heaven incident, the ultimate conviction was made possible because there was evidence.

3. Check the “AI Fraud Response” Clause in Your Insurance (Cost: 0 yen)

Confirm whether your insurance covers fraud or reputational damage caused by AI. Some cyber insurance policies now include coverage for impersonation and reputational damage response costs. In some cases, it can be added for a few thousand yen a month.

Simply ask your insurance agent, “Is there coverage for risks related to AI?” If you don’t ask, it won’t be suggested.

—

The Real Question is “Who Will Defend?”

To be honest, it is impossible for small and medium enterprises to tackle all attacks in the AI era alone. Even large corporations have not been able to completely stop 8.3 billion malicious ads.

That’s why it’s crucial to share a “system to lower defense costs” within the industry and community.

• Share information among peers about “what methods have emerged”
• Hold study sessions on AI fraud cases at local chambers of commerce
• Share monitoring tools and legal responses among multiple companies to distribute costs

The attackers have brought their costs close to zero with AI. The defenders must also find ways to lower their costs through systems.

A monitoring cost of 100,000 yen per month for one company can be reduced to 10,000 yen per month when shared among ten companies. This is a defensive strategy that can be implemented in local areas where small and medium enterprises gather. Large companies do it on their own. Small and medium enterprises fight by “banding together.”

—

Conclusion: Survival Strategies in an Era Where Attack Costs Approach Zero

AI is not only a convenient tool but also a tool that dramatically lowers the cost of attacks.

Insurance fraud, false complaints, malicious ads — in all these cases, the cost for the attackers is nearly zero. On the other hand, the damage for the victims can be in the hundreds of thousands of yen. This asymmetry will continue to widen.

For small and medium enterprises, the top priority is not to purchase expensive security tools. It is “a system to quickly detect anomalies,” “a habit of preserving evidence,” and “a structure to lower defense costs by banding together.”

All of these can be started today. Begin by registering your company name with Google Alerts. It takes just 5 minutes.