A recent report by NewsGuard, a website ratings provider, has revealed the emergence of a new wave of clickbait websites that use AI software to generate content. The study identified 49 websites across seven languages that appear to be created primarily or entirely by artificial intelligence language models designed to mimic human communication.
However, the report suggests that these websites are just the beginning and that there are many more websites that may be operating at a higher level of quality. Lorenzo Arvanitis, one of the researchers, acknowledged that their analysis may have overlooked some of these higher-quality, AI-generated sites.
The proliferation of these AI-driven websites poses several challenges for consumers and advertisers alike. As websites continue to grow, it becomes increasingly difficult to distinguish between human-generated and AI-generated content, which can lead to misinformation and inaccuracies.
Greg Sterling, the co-founder of Near Media, pointed out the dangers of AI-generated content, especially when it comes to health and financial advice. Advertisers are also at risk, as associating their brands with low-quality or misleading content can damage their reputation. Interestingly, some of these sites are reportedly using Google’s AdSense platform and AI Bard to create content, leading to an ironic situation.
Contrary to the assumption that AI-generated content is more objective, Vincent Raynauld, an associate professor at Emerson College, warns that developer bias is embedded in AI platforms, affecting results. The report suggests that consumers who rely on these sites for news may not necessarily perceive a drop in quality, as the content is often already considered inferior. AI-generated content allows website operators to reduce costs by eliminating the need for human writers, which may be attractive to some.
The report also raises concerns about the lack of transparency and ownership disclosure among these websites, which cover various topics such as politics, health, entertainment, finance, and technology. Some sites publish hundreds of articles daily, including content that promotes false narratives.
While AI software developers like OpenAI have implemented certain guardrails to prevent misuse, the researchers found that these tools can still be easily weaponized to produce misinformation and circumvent safeguards.
Identifying AI-generated content can be challenging without specialized tools, but the researchers discovered a common indicator among the identified websites. All 49 sites had published at least one article containing error messages commonly found in AI-generated texts, making it easier to identify their source. Concerns about the abuse of AI have sparked discussions about potential government regulation, although it remains unclear how effective such measures would be.
