Amazon Takes Aim at AI-Driven Advertising Waste
4 September 2026
Alan Chester – Chief Customer Officer
Amazon Ads has announced a new step in its efforts to tackle Made-for-Advertising (MFA) websites – low-quality sites created primarily to generate advertising impressions rather than provide genuine value to users.
It’s an issue that has existed in programmatic advertising for years, but the rise of generative AI is making it considerably more challenging. AI can now be used to generate content, fake author profiles and entire networks of seemingly legitimate websites at a speed and scale that would previously have been impossible.
Amazon says the same technology that can create a credible-looking blog can now be used to create hundreds of MFA sites in a single weekend.
Why does this matter to advertisers?
For brands investing in programmatic advertising, MFA sites can mean advertising budgets being spent on impressions that offer little genuine audience value.
And this isn’t a small issue. Research from the Association of National Advertisers continues to highlight the difference that media quality can make. Its Q1 2026 Programmatic Transparency Benchmark found that the best-performing advertisers converted 54% of their programmatic spend into qualified impressions, compared with just 32% for lower-performing advertisers. The research points to stronger measurement, tighter supply chains and greater control over media quality as major differentiators.
The challenge is also evolving. DoubleVerify recently uncovered a network of more than 200 MFA domains using automated AI-generated articles and imagery to create what appeared to be independent publishing sites, illustrating just how easy it has become to manufacture advertising inventory at scale.
Amazon is tracing the problem back to the source
Rather than simply identifying individual poor-quality websites, Amazon Ads is working with cybersecurity company HUMAN to identify the networks and operators behind them.
Amazon’s systems look at signals including ownership, monetisation paths, site structures and traffic patterns. HUMAN then independently analyses those networks across a much larger pool of internet traffic.
In one recent investigation, Amazon and HUMAN connected dozens of apparently unrelated websites through common ownership, identical site templates and shared legal information. They also identified fabricated author identities and AI-generated profile images. Traffic from the network increased fourfold between January and June 2026.
Amazon says its automated scanning and verification systems filtered out more than 10 trillion bid requests in 2025 that failed to meet its quality standards before advertisers could bid on them.
What does it mean for brands?
This is an important development for brands increasingly using Amazon DSP to reach audiences both on and beyond Amazon.
Programmatic advertising has always offered huge reach, but reach only matters when you’re reaching genuine audiences in environments that deliver value.
Amazon’s approach is increasingly about applying its data and technology not only to identify who advertisers should reach, but also ensuring greater confidence in where those impressions are being served.
As AI makes it easier to create poor-quality advertising inventory, we should expect supply quality and transparency to become an increasingly important part of measuring advertising performance.
For brands, the takeaway is straightforward: optimising media should no longer be just about CPM, reach or audience targeting – understanding the quality of the inventory your budget is buying matters too.