Imagine walking into a packed art museum in Buffalo on a cool October evening, not for an exhibition, but for a conversation about one of the most pressing threats of our time: the steady, engineered decay of truth. That’s exactly what happened at the Burchfield Penney Art Center, where Investigative Post gathered a sold-out crowd to hear Steven Brill, co-founder of NewsGuard and author of The Death of Truth, talk about why the internet has become what he calls a “toxic stew of disinformation.” Sitting beside him was Sofia Rubinson, a senior editor at NewsGuard, and together they painted a picture that was equal parts alarming and clarifying. The mess we’re living through online is not a random accident. It isn’t the result of a few bad actors slipping through the cracks. It is, in many ways, the very architecture of the modern digital economy. Bad actors—foreign provocateurs, domestic propagandists, partisan money machines, and now artificial intelligence—have figured out how to exploit the internet’s core incentive structure. Search engines and social media platforms don’t just tolerate their falsehoods; they willingly recirculate them because inflammatory content is the cheapest, most reliable fuel for traffic. “They realized that the more inflammatory your content was, the more eyeballs you would get, and therefore the more revenue you would get,” Brill told the audience. It’s a grim calculation, but once you see it, you can’t unsee it: the internet economy is not broken; it’s working exactly as designed for those who profit from division.
How did we get to a place where falsehoods outperform facts on the world’s biggest information platforms? Brill traced much of the blame to a legal and business framework that took shape in the internet’s early days. Section 230 of the Communications Decency Act, long celebrated as the law that made the open internet possible, gave digital platforms broad immunity from liability for what their users post. In principle, that protection was meant to encourage free expression and prevent platforms from being sued into oblivion over a single comment. In practice, it freed the biggest tech companies from any real responsibility for the content they amplified. Freed from legal pressure, Brill argued, platforms made a cold calculation: prioritize growth, prioritize advertising revenue, and let moderation take a back seat. “The business model of these companies—the Facebooks, the Twitters of the world—was to get as many eyeballs as possible because they were selling advertising,” he said. And what gets eyeballs? Anger. Outrage. Fear. The stuff that makes your heart race and your finger hover over the share button. It’s no coincidence that the most viral content often isn’t the most true; it’s the most charged. Then there’s the strange world of online advertising, where the old rules of media buying no longer apply. In television and print, advertisers generally choose the shows and publications they want to be associated with. Online, ads are placed based on data about you—your demographics, your habits, your location—regardless of which site you happen to be reading. That means a respectable company can end up funding propaganda without ever knowing it. Brill offered a jaw-dropping example: GEICO, the insurance giant owned by Warren Buffett’s Berkshire Hathaway, was for a time a top advertiser on Sputnik News, a Kremlin propaganda outlet. Why? Because Sputnik’s audience matched the demographics GEICO’s ad algorithm was targeting. The ad wasn’t placed because the company endorsed the content; it was placed because the data said you were looking. This is the hidden underbelly of programmatic advertising, where dollars flow not to quality but to clicks, and where disinformation becomes not just a social problem but a revenue generator.
For the past several years, NewsGuard has been trying to do something about this chaos. Since 2018, the organization has reviewed and rated roughly 36,000 online news and information platforms, scoring them on nine nonpartisan standards designed to measure accuracy, transparency, and credibility. Think of it like a nutrition label for news: a simple score from 1 to 100 that tells you whether a site is trustworthy or full of junk. The results are sobering. Of the 36,000 platforms reviewed, only about 12,000 received a passing grade. And of those that passed, just 10 percent earned a perfect score of 100. One of those rare perfect scores belongs to Investigative Post itself, a fact that drew a knowing nod from the audience. But the broader picture is troubling, especially with the rise of so-called “pink slime” news sites—partisan outlets financed by political action committees on both sides of the aisle that mimic local news in style but operate as propaganda in substance. These sites have been around for years, but Brill said artificial intelligence has supercharged them. AI models are trained on whatever is already on the internet, which means they absorb the falsehoods and stylistic tricks of these fake news outlets and reproduce them at scale. During the midterms, he said, these pink slime sites were doing the same thing that foreign propaganda operations have done: flooding the information ecosystem with stories designed not to inform, but to inflame. The hope, of course, is that rating systems like NewsGuard’s can help consumers navigate the swamp. But Brill made clear that the major platforms have not exactly welcomed the help. They were offered NewsGuard’s ratings and warning labels, and they balked. Why? Because sending users to reliable information would mean sending them away from the very content that makes the most money. As Brill put it, “The social media platforms are sending people to unreliable stuff, to hoaxes, to Russian propaganda … that wasn’t a bug. That was a feature.” The last thing they wanted, he said, was a little warning sign steering people away from the pure gold of outrage.
Given all this, you might expect Brill to be despairing. He’s not exactly optimistic, but he’s certainly determined. He acknowledged the intense blowback NewsGuard has faced for its work—political pressure from both sides, accusations of bias, and even congressional investigations. “We have faced lots of political pressure and lots of political opposition. We’ve spent hundreds of thousands of dollars on Congressional investigations, investigating us for being biased against conservative sites,” he said. He pushed back on that framing, insisting that “I take being non-political very seriously, but that doesn’t mean that people don’t get angry at us on both sides of the fence.” In a polarized age, that kind of centeredness is rare. What can the average person do? Brill’s advice is refreshingly practical. First, check the source. Don’t just read a headline and feel something; look at where the information is coming from, who funded it, and whether it meets basic standards of evidence. Second, diversify your media diet. If you only consume news from one outlet, or one algorithm, you’re seeing only a sliver of the world—a sliver chosen not for accuracy but for engagement. The more sources you read, the harder it is for any single propaganda operation to dominate your worldview. This is not glamorous work. It requires slowing down in a world designed to speed you up. But it’s also the only real defense against a system that profits when you stay angry and uninformed. Brill’s message was not that we should abandon the internet; it was that we should treat it with the same skepticism we would bring to any other powerful, unregulated force. Don’t trust the algorithm. Trust the brakes.
The newest frontier in this battle is artificial intelligence, and if you think the current internet is bad, Brill warned, AI is about to make it worse. Traditional chatbots like ChatGPT and Claude are trained on the entire internet, which means they’ve ingested vast quantities of lies, hoaxes, and propaganda. When you ask them a question, they’re not evaluating truth; they’re predicting patterns. As a result, Brill said, these tools are wrong about 30 percent of the time—an alarmingly high failure rate for something people increasingly rely on for information. He offered a chilling example: Russian platforms had generated false stories about Ukraine’s first lady, Olena Zelenska, spending American aid money on Cartier jewelry. The stories were fabricated from start to finish, but they spread across the web. Now, because AI models train indiscriminately, those fake stories are baked into their datasets. Ask a chatbot whether Zelenska shops at Cartier, and it might very well answer yes, citing only the made-up websites that planted the story. Meanwhile, the AI problem is compounding the existing “pink slime” problem, as partisan sites use generative tools to churn out content faster than ever. This is where NewsGuard is trying to offer an alternative. The organization’s latest product, NewsGuard AI, is built to do something radically different: it only draws from websites that NewsGuard has already verified as credible. Instead of scraping the entire mess of the internet, it relies on a curated layer of trustworthy sources. You use it like any other chatbot—type in a question, get an answer—but you also get something ordinary chatbots don’t provide: transparency. Rubinson explained that with NewsGuard AI, “you can know where every single piece of information that is on your screen is coming from, and you can read the original reporting to make sure that you’re fully understanding the issue.” That’s a subtle but profound difference. Traditional AI hides its sources behind a veil of confident prose. NewsGuard AI opens the hood and shows you the engine. And in another unusual move, it shares revenue with the publishers whose work it uses, helping sustain the journalism it relies on.
So is the problem of disinformation getting better or worse? Brill’s answer is both. On one hand, the forces fueling misinformation—AI, partisan money, algorithmic extremism—are more powerful and more sophisticated than ever. On the other hand, there are signs of countervailing momentum. Reputable news outlets that invest in real reporting, fact-checking, and editorial standards are finding ways to sustain themselves, and readers who care about quality are increasingly gravitating toward sources they can trust. The key, Brill said, is personal responsibility. “It’s an element of personal responsibility,” he told the audience. “I think we have to discipline ourselves to be aware.” That awareness is not just about avoiding obvious junk; it’s about recognizing that the information ecosystem we’re living in is not neutral. It was built by people with incentives, and those incentives often run counter to your interests. The good news is that you don’t have to be a media expert to protect yourself. You just have to be willing to pause, question, and verify. The future of journalism—and maybe the future of democracy itself—will be determined by whether enough of us are willing to do that work. In the meantime, organizations like NewsGuard and local reporting outlets like Investigative Post are trying to provide the tools and the accountability we need. For those in the Buffalo audience, there was one more reason to feel hopeful: they could try NewsGuard AI at a discounted rate thanks to Investigative Post, a small but tangible way to put a better information platform within reach. We may never fully escape the toxic stew, but with the right habits and the right tools, we can learn to cut through it.