02-06-2025
Navigating The New Ad Fraud Landscape: The Generative AI Challenge
Ashish Bhardwaj is a Engineering Lead at Google, building privacy preserving technologies to reshape the digital advertising.
getty
Anyone who's spent time in digital advertising knows it's a battlefield. For years, we've fought against ads that break the rules, clog up the works with spam or are outright fraudulent. It's a constant struggle to keep the ecosystem clean.
Now, generative AI (GenAI) has stormed onto the scene, and frankly, it's making our jobs a whole lot harder, at least for now.
GenAI is incredibly powerful. It can spin up slick, convincing ad content faster than ever before. The downside? Bad actors are using the exact same tools to create deceptive ads that are increasingly difficult to distinguish from the real.
Let's look into the primary threats, examine how GenAI is amplifying these challenges and, crucially, explore actionable strategies and technological advancements that business leaders in the adtech space can implement to mitigate these evolving risks.
To protect the integrity of digital advertising, we need to be crystal clear about what we're fighting. It boils down to three interconnected issues:
1. Ad Policy Violations: Think of these as breaking the rules of the road. It's a wide range, from ads making misleading claims about a product or promoting things they shouldn't (like sketchy pharmaceuticals or adult content) to technical fouls like using disruptive formats. Even things like ad size or how many requests are fired off fall under guidelines set by bodies like the IAB.
2. Ad Spam: We've all seen it: irrelevant, annoying, clickbaity junk. This isn't just about unwanted email anymore. In ads, it can be content designed purely to trick you into clicking (sensationalism) or technical spam like rapid-fire clicks generated without you even knowing. It degrades the user experience and makes people distrust all advertising.
3. Ad Fraud: This is where the real criminality lies—deliberate deception for financial gain. We're talking about fake clicks generated by bots or click farms, ads hidden from view but still counted as impressions (impression fraud) or faking valuable actions like purchases or sign-ups (conversion fraud). Fraudsters get sophisticated, too, mimicking legitimate websites to steal higher ad rates (domain spoofing) or secretly injecting extra ads on pages (ad injection).
It's crucial to understand that these aren't always separate problems; a fraudulent ad likely also violates policies and could certainly be considered spam. The tactics evolve constantly, meaning our defenses have to keep getting smarter.
GenAI is the ultimate double-edged sword in this fight. On the one hand, it's given the fraudsters powerful new weapons. On the other hand, it offers us new ways to defend the ecosystem.
AI can be used to create incredibly realistic deepfake videos for fake celebrity endorsements or elaborate scams. For example, deepfake videos of public figures—such as those created of Al Roker and Tom Cruise—can be used to promote bogus products or services.
Networks of over 200 AI-generated "slop sites" designed to mimic reputable publishers and defraud advertisers have been uncovered, filled with plagiarized or low-quality content to drive ad revenue. This is particularly frightening when combined with AI-powered bots that mimic human browsing to generate fake traffic and clicks, like the CycloneBot scheme targeting connected TV platforms by inflating views.
Thankfully, the good guys have AI, too.
AI, for instance, is improving the ability to prevent fraudsters and keep billions of policy-violating ads from ever showing, as Google research shows. Companies are using advanced machine learning to spot and block fraud in real time, often before it causes real damage.
These algorithms learn and adapt, getting better at recognizing new fraud tactics as they emerge. We can even use AI to create synthetic data to train our fraud detection models, making them even sharper without using real user data.
GenAI tools are becoming widely available, often through open-source platforms. Generative AI isn't yet a game-changer for novel attacks, but it allows threat actors to move faster and at higher volume. This means even less technically savvy crooks can now generate convincing fake content and automate their scams.
It's democratized fraud creation, and I anticipate a significant spike in AI-driven risky ads because of it.
The challenge of risky ads, amplified by GenAI, is real, complex and evolving. However, proactive measures can be taken. Business leaders in advertising and ad tech should consider the following:
• Invest in advanced detection and verification. Companies should invest in AI-powered fraud detection tools that go beyond traditional rule-based systems. This includes solutions for detecting AI-generated content, manipulated media (deepfakes) and bot traffic.
• Promote transparency and ethical AI use. Advocate for and adopt transparent practices in AI development and deployment. This includes clear labeling of AI-generated content and adhering to ethical guidelines to prevent bias and misuse. The IAB's "Generative AI Playbook For Advertising," for example, offers guidance on practical applications and ethical considerations.
• Foster collaboration and information sharing. The adtech industry should continue to work collaboratively to share information about new threats and effective countermeasures. Industry bodies can play a key role in establishing standards and facilitating this exchange.
• Focus on high-quality, human-verified inventory. Prioritize advertising on platforms and with publishers that demonstrate a commitment to combating fraud and maintaining high content standards. Consider solutions that help identify and avoid low-quality sites.
• Adapt and innovate continuously. Companies must foster a culture of continuous learning and adaptation, staying on top of new GenAI capabilities and being prepared to adjust strategies and technologies accordingly. This includes exploring emerging approaches like smaller, more efficient AI models (SLMs) and advanced data provenance techniques.
While GenAI has thrown gasoline on the fire of ad fraud, it also provides tools we need to fight back more effectively. By understanding the risks, investing in the right technologies and fostering a collaborative industry approach, we can navigate this new battlefield and work towards a cleaner, more trustworthy digital advertising ecosystem.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
