As we see a spike in low-quality AI-generated content across the internet, media measurement and verification companies have an opportunity to help advertisers navigate the surge.
On Monday, DoubleVerify launched a GenAI Website Avoidance & Detection offering as part of its technology suite. The tool helps advertisers identify and block low-quality, AI-generated content pre-bid and measure its presence post-bid. The rationale: these sites harm both campaign performance and brand reputation.
The company described the tech as “nuanced,” evaluating sites and subdomains based on a few factors, including “repetitive” cookie-cutter content formats, chatbot-generated text in articles (which often involve telltale phrases), wonky or leaked content placeholders, among other signals. DV makes it clear that not all AI content is bad, of course, and that they’re specifically looking for “low-quality” AI content based on the markers described.
Why This Matters:
GenAI tools like ChatGPT, Claude, Grok, Perplexity, and Gemini have made it incredibly easy to create large amounts of content in seconds. While this initially applied to text, advancements in AI-generated imagery and video (e.g., OpenAI’s Sora) are accelerating. As this happens, we’ve seen an explosion in MFA (made-for-advertising) sites that rely on AI to produce at scale. DV already has an MFA offering—this seems like it will help them dig in more on another dimension of the problem, targeting junky AI sites more deeply.
According to the company, they have already “identified numerous long-tail websites that leverage GenAI tools to churn out low-quality content, often rife with errors, editorial inconsistencies, and plagiarism.”
Brands, of course, are left to grapple with this (growing) issue. Inclusion lists are gaining traction as a response but can be hard to update and may suffer from scalability issues. Exclusion lists face even more challenges—they are only as effective as what is known to exclude, and AI sites can scale rapidly, especially under sophisticated network owners/bad actors.
Experts React:
AI detection company Copyleaks issued a study earlier this year showing an 8,362% (!) surge in AI-generated content online between November 2022, when ChatGPT-3.5 launched, and March 2024.
Co-founder and CEO Alon Yamin said, “The demand for scalable and cost-effective content creation across industries like e-commerce, marketing, and media has further propelled the adoption of AI-generated content, driving its exponential growth on the web.”
Hmm, we have MFA — maybe we call this “made-with-AI” content?
Our Take:
AI content on the web is a constant game of whack-a-mole for advertisers, platforms, and measurement and verification companies. It’s very reminiscent of cybersecurity from a threat detection standpoint. Tomorrow the technology gets better, and bad actors get smarter. AI-generated video technology is also rapidly advancing, creating even more complexity around the problem.
It will be interesting to see how the industry keeps pace with protection. It also stands to reason that “waste” here, which is impossible to stop entirely (just as no company is 100% safe from a cybersecurity standpoint), will grow larger and larger moving forward and be baked into costs.
