The old ad tech adage is that the bad guys follow the ad spend stream, but will the same apply to generative AI soon?
The popularity of large-scale language models has led AI to create massive amounts of text, image, and video content, so advertisers may unwittingly fund low-quality content. There is an increasing focus on the question of whether
A new report shows how questionable websites are quickly publishing and monetizing AI-generated content. Earlier this week, researchers at news credibility service NewsGuard found out how hundreds of programmatic ads paid by blue-chip brands were delivered to a growing number of AI-generated websites churning out hundreds of articles every day. announced a detailed study of Day.
Over the past two months, the research team found nearly 400 advertisements for 141 major brands on over 50 websites while browsing the Internet in Germany, France, Italy and the United States. But unlike other recent NewsGuard reports on new types of AI content, the latest findings weren’t necessarily misleading. Instead, low-level publications ranged from plagiarized versions of actual news stories published elsewhere to clickbait headlines promoting unproven or potentially harmful treatments for allergies, ADHD, and even cancer. Found quality content. NewsGuard’s list of AI-generated “unreliable” websites also appears to grow rapidly, jumping from just a few dozen in May to over 200 in June.
Jack Brewster, editor of NewsGuard Enterprise, told Digiday, “The creation of credible AI-generated news sites is being driven by the monetization of large ad tech companies that are monetizing these sites en masse.” Told. “and [they] It doesn’t appear to have human oversight or check for correctness. “
The brands were likely unaware that their ads were running on AI-generated websites, so NewsGuard decided not to name the advertisers. However, examples range from big banks and streaming services, to tech companies and auto giants, to sportswear and pet suppliers. Of the ads identified by NewsGuard, over 90% were delivered via his Google Ads.
“These companies aren’t directly saying, ‘Can we object to this AI-generated news site?'” Brewster said. “They just tell Google or another third party to serve ads to people like you and me, and that creates another problem.”
Advertisers’ AI-related brand safety concerns are already generating new business for companies like DoubleVerify as companies look for new ways to stay safe. The company announced last month that its AI content farm increased its brand safety technology by 56% in Q1 2023 compared to 2022.
AI-generated content isn’t all that different from other brand safety concerns, but DoubleVerify CEO Mark Zagorsky said that along with emerging issues such as concerns related to piracy, AI is He said that the scale of the generated content is creating new challenges. As a result, more and more advertisers are adding his AI-generated website to their blocklists. Other advertisers worry less about AI-generated content, and instead he focuses more on AI-generated content. DoubleVerify is also stepping up its investment in his AI tools. The company’s first-quarter 2023 earnings show that product development costs increased to $28.5 million from $21.5 million in the same period last year. (Zagorski said the upgrade will help develop new ways to detect content across more languages and more content formats, including video.)
“The interesting thing is whether or not this was created by generative AI is a more important factor than what the content itself is,” Zagorsky told Digiday. “That’s why I want to use a scalpel instead of a kitchen knife.”
Evelyn Mitchell Wolfe, Senior Analyst, Digital Advertising and Media Analyst at eMarketer, notes that generative AI adds new challenges to the programmatic advertising ecosystem while exacerbating existing weaknesses. increase. The challenge is for traditional publishers struggling between using generative AI tools, investing in human-generated content, and giving AI models API access to high-quality content to use as training data. It poses an existential crisis for He added that exclusion lists do not guarantee that advertisers can avoid all dangerous content.
“Generative AI dramatically increases the surface area where low-quality content can exist,” Mitchellwolf said. “The problem is snowballing.”
Asked to comment on the NewsGuard report, Google spokesman Michael Aciman said the company had scrutinized the AI-generated websites mentioned in the NewsGuard report and found that many “because of widespread policy violations.” It said it had removed ads from its website.Google has stopped monetizing some other websites Individual pages on sites that violate Google’s policies, cited by NewsGuard. Aciman also said that just because a website has AI-generated content doesn’t necessarily mean it’s against Google’s policy, but he’s aware that “bad guys are always changing their approach.” added.
“We have strict policies governing the types of content that can be monetized on our platform,” Aciman said. “For example, we do not allow ads to run alongside harmful, spammy or low-value content, or content that is simply copied from another site. When applying these policies. focuses on the quality of content rather than how it is created, and blocks or removes ad serving when violations are detected.”
These challenges are emerging as more attention is focused on other parts of the programmatic advertising ecosystem. A new study on the programmatic media supply chain shows that “made for advertising” (MFA) websites account for 21% of impressions and 15% of total ad spend. The report, published this month by the National Association of Advertisers, also found that MFA websites account for 19% of open marketplace media purchases and 14% of private marketplace transactions.
MFA websites include many different types of websites besides those with AI-generated content, but research shows that advertisers don’t always control their ads. indicates that The report also shows how much more can be done to enable advertisers to fund quality content rather than clickbait from both humans and bots.
Keri Bruce, an attorney at law firm Reed Smith, which produced the ANA report, said AI made it easier for websites to be created significantly faster, making brand suitability more difficult and “bad guys.” ‘ will be able to get more profit. All of this will lead to a greater “legal whack-a-mole,” she said, and advertisers should keep track of how many websites they operate and put more emphasis on inclusion lists rather than just exclusion lists. added.
“I can’t name the 44,000 websites I visit, and I don’t think one consumer can do that,” she said. “That’s the programmatic challenge. You can advertise on thousands of websites, but do you really need to? Are you visiting thousands of her websites?”
