Google’s AI Is Spitting Out Lies - And You’re Falling for It

We’ve all done it - typed a quick question into Google and trusted the first thing we see. Whether it’s looking up a recipe, checking a fact, or bless our hearts, searching for medical advice, we rely on search engines to deliver accurate answers instantly. But what happens when those answers are dangerously wrong? Worse yet, what if we never realise it?

Recently, Google’s AI-powered search feature has been caught delivering outright false and provenly harmful information. Users were horrified to discover it recommending people eat rocks, add glue to their pizza sauce, and disturbingly, suggesting a dangerous recipe for chlorine gas when someone searched for ways to clean their washing machine. The AI, designed to summarise search results, is pulling information from satirical websites, joke comments on Reddit, and some outright falsehoods - to present as fact.

The Death of Critical Thinking in Search

The way we interact with search engines has changed. People no longer scroll through multiple sources, cross-checking information. The first result - especially when formatted as an AI-generated answer - feels like the final word. Google knows this. That’s why AI Overviews sit at the top of search results, framed as the most relevant and reliable information.

But here’s the danger: if AI-generated summaries are wrong, how many people will notice? Most users won’t click through to verify. Many won’t even read past the summary. This blind trust in search engine AI makes misinformation more powerful - and more dangerous - than ever before.

When AI Lies, the Consequences Are Real

The implications of these errors are terrifying. Imagine an individual looking for a simple cleaning tip, only to be met with a deadly chemical reaction. Or a student using AI-generated search summaries for research, unknowingly citing and perpetuating misinformation. AI hallucinations aren’t harmless glitches; they have real-world consequences.

Companies like Google assure us that these errors are rare. But the reality is, generative AI models have a well-documented history of “hallucinating” - fabricating information with complete confidence. “Rare” is no longer rare when you have this many people using it that many times a day.

Why Is This Happening?

AI models don’t “know” anything. They don’t verify facts or cross-check information the way a researcher or journalist would. Instead, they predict the most probable sequence of words based on their training data. If that data is flawed - or if the AI pulls from unreliable sources - it has no mechanism to correct itself.

Unlike humans, AI lacks common sense, morality, and the ability to question its own outputs. Once an AI-generated summary appears in a search result, it gains credibility, simply because it came from Google. And once misinformation spreads, it’s nearly impossible to undo.

The Threat to Research and Academia

For researchers and academics, the spread of AI-generated falsehoods is particularly alarming. Reliable information is the foundation of scientific inquiry, yet AI’s tendency to hallucinate could distort findings, mislead scholars, and weaken the credibility of academic work. If researchers unknowingly cite AI-generated misinformation, it can ripple through the academic world, affecting the integrity of future studies.

Academic institutions and researchers must take a proactive approach to verifying sources, ensuring they rely on well-documented, peer-reviewed materials rather than AI-generated summaries. The convenience of AI-powered search cannot come at the cost of research credibility. As misinformation continues to spread, it is more important than ever to foster a culture of critical thinking and rigorous fact-checking within academia.

The Illusion of Authority

Search engines, once a gateway to information, are increasingly acting as gatekeepers - deciding what is seen, what is prioritised, and what is believed. AI-generated overviews further erode our ability to critically evaluate information by presenting a single, often flawed, response as the definitive answer.

If we don’t start questioning these results, misinformation will become embedded in public knowledge, shaping opinions, influencing decisions, and even affecting policy. The responsibility doesn’t just lie with AI developers - it’s on all of us to remain sceptical, verify sources, and resist the temptation to accept AI-generated content as truth.

The internet was built on the promise of open access to knowledge. If we don’t challenge the rise of unchecked AI-driven misinformation, we risk replacing that promise with a dangerous illusion of authority - one where truth is no longer determined by facts, but by algorithms.