[DECODED] Google is changing your search habit. Here are 3 things to keep in mind.

Do you find Google’s AI overviews helpful — the summary that appears at the top of your search page whenever you look something up on Google? It used to show top results that led to distinct web pages. It used to be so simple, until it got replaced with paragraphs that synthesize potential answers to your query.

Since 2023, Google has been rolling out generative AI features on Google Search. It launched the search generative experience in May 2023, available only to users who opt in via Search Labs. In 2024, it rebranded SGE as AI overviews and successively rolled it out to over 100 countries, including the Philippines. The rollout did not require an opt-in from users. Then in 2025, it launched AI mode, which lets users interact with its Gemini models like a chatbot while conducting a simple Google search.

In May, Google launched the biggest overhaul of its search engine, calling it “a new era for AI search.” It comes with a new model that is now capable of agentic AI assistance, which means the “search” results can now also include task execution, such as booking reservations or even contacting businesses on the user’s behalf.

These features are now available on Google Chrome. While some users have reportedly switched browsers in response, there are still some who might find these updates unavoidable. So if you ever decide to explore these features, here are three things you should keep in mind.

1. Ease can come at the cost of privacy risks.

Google’s latest update includes “personal intelligence,” which allows users to connect apps like Gmail and Google Photos so that search results are tailored to each user’s personal context.

Granted that these already live within Google’s ecosystem and your personal data don’t necessarily get handed over to a third party. But connecting it to Search gives Google a more unified view of you than ever. Information about you that was somewhat kept in separate apps is now explicitly combined to let one company decide what you need and what you don’t, increasing reliance on it.

1. News publishers are losing as AI takes over search.

The first iterations of Google’s AI features in search completely disregarded the publishers whose content it uses to train its AI models. Google has since added links to its AI summaries in response to the backlash, but publishers continue to bear the brunt of these tech developments.

A study from Pew Research Center found that users tend not to click on the links within AI summaries, with only 1% of visits to the source material, hurting traffic to the very publishers whose work made those summaries possible in the first place.

Particularly in the Philippines, this creates more friction for independent news groups that are fighting for attention in a space already cluttered with content creators who operate with little to no regard for journalistic standards or verification.

READ: [DECODED] Franco Mabanta and PGMN’s business model

Facts may continue to suffer, too. At its core, AI works as a pattern recognition at scale. It is trained on big data, and what it essentially does to answer queries is give the most probable prediction based on the patterns it inferred from all of the data it was fed. 

So the reason it makes mistakes is that it operates on probabilities, not on the understanding of truth and facts.

This is particularly a cause for concern in the Philippines, where disinformation is quickly spread online. For instance, our Decoded story last week discussed how false charter change narratives spread online amid the ongoing Senate chaos. These narratives proliferate because it has become increasingly difficult to identify what is factual in an already cluttered online environment, and platforms do very little to temper their spread.

READ: [DECODED] How charter change narratives spread online amid Senate chaos

1. It’s hard to hold anyone accountable when AI gets things wrong.

In 2024, screenshots of inaccurate AI summaries that told people to eat rocks and put glue on pizza infamously circulated, causing Google to restrict its AI search tool. But it did not derail any of the company’s broader AI ambitions, even though inaccuracies continued to show.

In January 2026, The Guardian reported that Google’s AI summaries showed inaccurate health information about liver function tests. Instead of taking accountability, The Guardian said that Google simply removed the summaries for the search terms related to liver tests.

READ: AI vomits stupid things. And we catch them with our mouths wide open.

This happens when the technology is moving faster than the accountability structures designed to govern it. Since the rise of social media, we’ve been calling for accountability from the platforms for allowing the spread of disinformation and hate. Now, with AI, the problem has compounded.

I’m not against AI use. I think it has evolved so fast that it’s now a matter of living with it and using it responsibly. And AI, for its worth, actually does have a lot of useful applications, such as in medicine, education, and even journalism and community building.

This is why The Nerve and Rappler are conducting AI workshops that help people from different sectors navigate AI thoughtfully and effectively, and apply it in real-world settings. If you’re interested, go here to learn more. – Rappler.com