The Sift: Are AI summaries accurate?

April 13, 2026

Artificial intelligence Journalism Misinformation

An educator's guide
to news literacy

April 13, 2026

Hi there!

Interested in winning a $100 Amazon gift card? We are giving away 15 of them!

Share your opinions on The Sift in this survey, and at the end of it you can enter to win a gift card.

We aim to make The Sift helpful to educators and are always looking for ways to improve the newsletter and all of our resource offerings, so please don’t be shy about sharing your thoughts.

Thank you!

— The Sift team

In this issue

Are AI summaries accurate? | RumorGuard slides

Daily Do Now slides

A week’s worth of do-nows, just for you.

Top picks

Here are the latest news literacy topics and tips on how to integrate them into your classroom.

A computer screen shows a search bar and search results in gray boxes.

Google AI summaries aren’t always accurate.

1. How accurate is Google’s AI Overview?

Type in a topic to search on Google, and the first result you’re likely to see is an AI-generated summary. Can you trust it?

Many inaccuracies: A recent analysis found that these summaries are accurate 91% of the time, which might sound fine at first. But considering the high volume of Google searches — five trillion a year — it adds up to hundreds of thousands of inaccurate summaries generated every minute.
Errors acknowledged: Google acknowledges that its AI Overview “can make mistakes” and advises users to “double-check responses.”

💬 Discuss:

What experiences have you had with AI search summaries?
When should you verify the information provided by AI tools, such as search summaries?
If AI summarizes credible sources, what will happen to the web traffic on those websites? Will they be able to survive without people visiting their site?

💡 Idea: Create a contest to see which student(s) can catch Google’s AI giving false information in its summary first — or the most. Then discuss the experience as a class. How often, overall, were AI summaries right or wrong? What might the effects of an increasing number of people relying on AI summaries be?

⭐ NLP Resources:

Videos: “AI Overviews has a misinformation problem” and “Here’s how to search smarter on Google” (NLP’s YouTube page)
Activity: “Fact-check it!”

🔗 Related:

“Minnesota media professionals say Google AI Overview poses threat to news” (The Minnesota Daily, student newspaper at the University of Minnesota)

2. Exploring the ethics of sharing misinformation

Social media platforms are designed to make sharing and resharing quick and easy – but pausing to ponder whether it is ethical to share something may help reduce the spread of misinformation.

Sharing misinformation: Social psychology researchers say people generally know that sharing misinformation is ethically wrong, but often do it anyway.
Considering ethics: A study in the U.S. found that when people were given fake stories (some of which aligned with their political beliefs) and asked to list reasons why sharing them was either ethical or unethical, they were much less likely to say they would share them. The study compared this group with two others, one that rated ethics without giving reasons and one that wasn’t prompted to consider ethics.

💬 Discuss:

What might cause people to knowingly share misinformation?
Have you ever shared something you either knew or suspected was false? Perhaps because you believed in the underlying message?
If sharing misinformation is an unethical act, is sharing high-quality, credible information an honorable one?

💡 Idea: Use the “Reflect” slide in Week 21 of the Daily Do Now resource to further reflect on this topic.

⭐ NLP Resources:

A News Literacy Project infographic says, "Impostor alert: Don't fall for fake Labubus or news knockoffs" and shows several images of Labubus.

Infographic: “Impostor alert: Don’t fall for fake Labubus or news knockoffs”
“Misinformation” (Checkology® virtual classroom)

3. From TikToker to news creator

One of the most popular news creators today is Vitus “V” Spehar, who first went viral when explaining the news from under their desk during the Jan. 6, 2021 Capitol attack.

Creator journalist: Spehar started as a self-identified TikToker, but has come to consider themself as a journalist (link warning: explicit language). They have built an audience of millions posting Under the Desk News videos about U.S. politics and culture and publishing a Substack newsletter.
A mix of sources: Spehar says they gather information from legacy news sources and tips from viewers and readers. They sometimes verify details directly with politicians’ offices or primary sources like the U.S. Supreme Court docket.
Creator or journalist? News creators (people who make content about news and sometimes do original reporting) have become a popular source of information, making news more accessible and entertaining for audiences. Spehar suggests a third space for news creators to work with journalists. “There’s room enough for all of us and together we get the job of telling the truth and holding power accountable done,” they said.

💬 Discuss:

What makes someone a journalist? What is the difference between journalists and most talk show hosts? Or influencers online who talk about news?
Which kinds of creators try to follow standards or ethical guidelines to ensure accuracy and fairness? When are these standards important? Are they ever not needed?
What’s the difference between actual news reporting and content about news? Could content about news survive without original reporting?

🗒️ Note: Check out NLP’s “Teach about the freedom of the press” page for more resources.

⭐ NLP Resources:

Poster: “The Information Ecosystem Depends on News”
Infographic: “Seven standards of quality journalism”

🔗 Related:

“The Complicated Intersection of Digital Creators and Journalists” (Ebony)

These interactive slide decks help students think carefully and critically about viral claims in the digital age.

No, the Artemis II historic lunar mission isn’t being staged

A post reads, “THE TRUTH WILL GIVE YOU FREE” and features an image that appears to show the Artemis II astronauts filming part of their mission against a green screen. The News Literacy Project has added a label that says “NOT STAGED.”

❌ NO: This is not a genuine photograph of the Artemis II astronauts faking their lunar mission by filming it on a green screen.

✅ YES: This image is AI-generated and was created by someone unaffiliated with NASA.

✅ YES: Additional rumors are circulating online falsely claiming that the Artemis II mission was staged or faked in some way, including one involving the flight’s plushie mascot.

✅ YES: There is overwhelming evidence that four astronauts flew around the moon aboard the Artemis II spacecraft this month.

⭐ NewsLit takeaway:

AI tools can create realistic visuals that appear convincing at first glance. To muddy the waters even further, the widespread prevalence of these fabrications has led some people to doubt genuine videos and photographs. Check out these Google Slides for a few tips to discern space facts from fictions.

Video doesn’t show Hidden Valley Ranch logo in real life

❌ NO: The label shown on this bottle of Hidden Valley Ranch is not the genuine label.

✅ YES: This label was digitally altered to resemble the valley in the video.

✅ YES: The viral video shows a valley in Marin County, California.

⭐ NewsLit takeaway:

When we want something to be true — whether it’s a claim confirming a political opinion, a rumor about a celebrity or an amusing anecdote about a bottle of ranch dressing — we are less inclined to investigate and check the facts. Correcting this bad habit may not seem all that important when stakes are low, but our opinions can become distorted by misinformation when we repeatedly accept false claims as the truth. Check out these Google Slides for tips on debunking these feel-good claims.

➕ An American journalist kidnapped from a street corner in Baghdad was released last week by the Iran-backed Iraqi militia that had been holding her. (For discussion questions and teaching ideas on this topic, check out the last Sift issue.)

➕ Students may be the future of journalism … but they’re also the present. Student journalists in the U.S. are on track to produce a quarter million bylines in campus newspapers this year.

➕ Itchy eyes are one of the alleged symptoms of bixonimania, a nonexistent disease made up by a medical researcher in Sweden who wanted to see if AI chatbots would share this misinformation as medical advice. (Spoiler: They did.)

➕ Lego animated movies are popular among children — and with pro-Iran groups that are using AI to produce propaganda videos featuring Lego political characters.

➕ AI tools are an almost inescapable reality for Gen Z, but a recent study found that they’re angry (31%) and anxious (42%) about the technology.

➕ That’s not a glitch, it’s a dial tone. A mother recently explained this to her 9-year-old daughter after she got her first landline phone, part of a growing trend for families trying to reduce screen time.

➕ While there is optimism that nonprofit newsrooms can help replace the loss of local newspapers across the country, they only produce a fraction of the revenue they need to survive and rely heavily on philanthropic support.