This Week’s Fact Check (July 25-31, 2026) – Cyberattacks & Destroying Books

This week, I’m tackling Trump blaming Minnesota for the cyberattacks on their water system and be careful of which photos of world disasters you believe this week; Google, for one day, allowed people to generate satellite images with Google Earth with a click of a button. That didn’t last long before it was woefully abused.

And I’ll touch briefly on Anthropic’s Project Panama and the great book destruction crisis that’s been all over social media this week, but I am going to keep it light here. (ETA: I optimistically wrote that I’d keep it light before I inevitably dove deep into it below.)

Trump Blamed Minnesota for the Cyberattacks on Their Water System

Sunday through Monday, cyberattacks affects the water in more than 30 cities across Minnesota, causing the water to be shut off for a period of time while teams investigated and contained the attack.

Just a week earlier, federal agencies sent out warnings that Iranian hacking groups, such as CyberAv3ngers, were attempting to target internet-connected operational technology devices. (1)

“I blame it on Minnesota because they’re grossly incompetent,” Trump said at the Camp David presidential retreat in Maryland. “I would blame it on Minnesota and the governor, the corrupt governor of Minnesota. Iran’s got bigger problems than worrying about Minnesota.”

However, at least six other states have reported cyberattacks or attempts recently, including Wisconsin. (2)

Meanwhile, social media took to spreading their own theories as fact, citing as many sources as the President did.

The truth?

At this moment, there are no clear answers.

Intelligence agencies have assessed that Tehran was likely behind a coordinated attack on more than 30 municipal water systems in Minnesota.

Iran is a convenient scapegoat, with motive, and a pattern of behavior, since there were reports of Iran hacking into water and oil systems in the US back in April. And if we still believe our intelligence agencies, then there you go.

Google Allowed Generative AI Satellite Images for a Day, Pulled It For “Deepfake” Issues

You can see the announcement including the update on Google’s site.

This isn’t exactly a fact check as much as a heads up.

On July 30, 2026, Google announced the ability to use their Nano Banana to generate “custom images using Google Earth’s satellite”.

Immediately, that went how you might expect.

NPR reported that Google pulled the feature the next day after people were reportedly using it to create “deepfake” style imagery of places around the world being bombed, presumably to post on social media in order to spread disinformation.

Evan Hill@evanhill

So it seems Google activated its AI image generation program inside its satellite image app today (blog post here: blog.google/products-and-p…). Very curious how or if this idea was red teamed internally because the opportunities for abuse and disinfo are literally boundless.

𝚑𝚎𝚗𝚔 𝚟𝚊𝚗 𝚎𝚜𝚜 @henkvaness

Tonight I typed just one sentence into Google Earth and put refugees near the Mexican border. Then I planted a nuclear plant in Iran. What on earth is Google doing? Check my latest post here: https://t.co/N0h6CVECqS

1:31 PM · Jul 30, 2026 · 43.5K Views

---

17 Replies · 174 Reposts · 580 Likes

So as always, if you see inflammatory imagery of locations around the world being bombed, double check the source and that it’s not disinformation.

Anthropic is Destroying Rare Books, or Are They? And Can They?

This story took off this week, but the actual story is at least five months old.

There was a flurry of information going around, and since AI is a very polarizing topic right now, people were getting loud about it, on both sides of the line, for and against generative AI.

Here are just the facts based on the court filings and the original Washington Post articles that uncovered them: (4)(5)(6)

• Anthropic is paying $1.5 billion to settle a class-action lawsuit brought by book publishers and authors for violating copyright by downloading the books without permission.
• The company has since spent tens of millions on millions of physical books in order to scan the books to train their AI, in a move approved by the judge.
• Anthropic’s source for physical books included buying from book retailers Better World Books and World of Books.
• Anthropic is slicing the spines off the books in order to more quickly scan the inside pages.
• There are other less destructive ways to scan a book.
• There are known and utilized methods to obtain scans of books legally that involve pursuing obtaining licensing of the books to have permission to scan them and use them. Reportedly, AI companies have looked for ways to get around this process.
• The scanning of pages is to feed knowledge into the AI models behind Anthropic’s products, including its chatbot, Claude.
• A judge ruled this fair use in 2025, after Anthropic was caught pirating books via downloading them using torrents. As long as the books are “legally obtained”, Anthropic can scan them.
• In the court documents, Anthropic does not claim to be scanning the books to “preserve”, as many have claimed on social media.
• Claimed use for the books include: training, research, development, and production of AI models and associated products and services.
• Reportedly, Anthropic chose books pre-2022 to meet their goal of training their AI to write more like a human would, and books pre-2022 would not contain any genAI-written text.
• An internal planning document stated that “Project Panama is our effort to destructively scan all the books in the world”.(7)
• The same document instructed staff to keep quiet, adding that “we don’t want it to be known that we are working on this’”.(8)
• Anthropic is not the only one training their AI on books; Meta, Google, Microsoft, and OpenAI are also facing copyright lawsuits from book authors.
• Meta has also downloaded millions of books for free from torrent platforms rather than paying for the books in a legal way, and is facing lawsuits because of this.
• Anthropic’s approach to physically purchasing the books to scan them has been called “a good example of the company taking a more restrained approach and achieving legal compliance.”
• Anthropic’s approach to scanning the books does not generally qualify as “digitally preserving” them because that implies granting access to the scanned copies. They are being used to train their models, but any distribution of the books would be illegal without the authors’ permission.(10)
• The destruction of books is a judge-approved move to avoid further copyright claims that digitally scanning them is creating a copy of the book, which would be illegal. It is called “destructive scanning”, a move out of caution, considering they have already faced lawsuits over copyright.
• In 2025, OpenAI and Microsoft announced they were working with Harvard’s libraries to train AI models on nearly 1 million public domain books dating back to the 15th century, fully digitized without their original copies being destroyed. So Anthropic could have found a similar method to train their AI models that did not require destruction of millions of books. (9)

So Are They Destroying Rare Books?

From my research, I have seen no credible claims that Anthropic sought out specifically “rare” books outside of the books being older than 2022, but there is also no evidence that Anthropic was careful to not destroy books that have fewer copies in circulation. There is more evidence that points to Anthropic simply purchasing en masse, with less consideration of the effects, especially considering the fact that they had no problem pirating books illegally previously and that they didn’t want Project Panama to be known to the public.

Does that absolve them of criticism for what they’re doing? I don’t believe so.

I would like to point out here, since it’s my blog, how I saw the conversation transpiring on social media.

A lot of people were upset that AI and the books topic seemed to be so polarizing and that the anti-genAI people, in particular, were unwilling to “accept that there could be nuance”.

This is a bit of a false argument, if you zoom out a touch. The people in the anti-genAI camp understand that there are more details to this story than just that Anthropic is consuming and destroying old books to train its models on. And that’s the problem.

We have been inundated with news stories about AI data centers polluting minority and impoverished neighborhoods, AI hacking other companies, AI generating CSAM and Republicans inexplicably defending it, and the list of grievances goes on, and there is still no clear concerted effort to create regulations for AI.

So when the news broke about Project Panama’s “efforts to scan and destroy all the books in the world”, for a lot of us, it wasn’t that we only cared about the books, but it was more like, “and now the books, too??“

And the destruction was fueled by company greed and unnecessary.

So yeah, people who enjoy books are a little upset.

A New Pro-AI Talking Point: “It’s Not Black and White”

Just a note: I monitor social media for repeated and trending talking points from multiple sources, because it can be a sign of misinformation or disinformation.

The new talking point I’ve seen pop up includes comments like, “AI isn’t black and white. You’re ignoring that there’s nuance.” “I just have a nuanced take about AI.”

This is a logical fallacy, because it assumes that those who are against using generative AI don’t accept that AI can provide some good uses.

Nuance or lack thereof isn’t what’s at issue for those who are against generative AI.

You can be against the use of something, and still accept that there are potentially good things that can come from it.

But in the case of many people who are against generative AI, the copyright theft, the CSAM generating, the data centers polluting impoverished neighborhoods, the lack of regulations, the studies that show negative cognitive effects, etc. outweigh the fact that AI can help you finish your work faster.

So someone who is against AI can choose to be against it, even if it does good things, because they don’t think the benefits of use outweigh the costs.

That is not black and white thinking or ignoring nuance; it’s understanding that things that can create some good can also create more harm than they are worth, and some people choose to not participate in something that they believe creates that level of harm.

AI Misinfo Check

Maybe I’m going to do this more often, as a little side bar, because it comes up with every topic I dig into. In my research, I inevitably use Google, which inevitably spits out its Gemini AI summaries. I always scroll past, but sometimes what it shares is jaw dropping. Statements in this screenshot aside, look at the main sources listed. Do you see the glaring problem?

Instagram and Youtube should not be the primary sources listed for this topic. This story was covered in depth by reputable news sources who did the investigation into the court cases and what was going on. I list some of them in my own sources. Social media regurgitates what it finds, often just from skimming headlines and not actually reading the full articles.

The Times of India, by the way, is not a reputable source. It’s one of the RSS news aggregator sites that is commonly used by disinfo campaigns to spread their false information. Do not trust that site as a source

I don’t know how I can scream any louder about how basic use of LLMs is so dangerous when it comes to spreading misinformation, but I’ll keep sharing the examples I come across in the meantime.

Thanks for reading. I do my best to verify all the information, but if I got anything wrong, let me know. I write weekly fact checks as well as other articles on misinformation and disinformation happening on social media and beyond, so if that’s your thing, please subscribe. If you find my work valuable, please consider a paid subscription to help support my work.

Sources

1. Wood, Colin. (2026, July 28). “Coordinated cyberattack disrupts water utilities in 30+ Minnesota communities”. Statescoop. https://statescoop.com/coordinated-cyberattack-disrupts-water-utilities-in-30-minnesota-communities/
2. Lyngaas, Sean. (2026, July 31). “Sweeping cyberattack on water systems in multiple states has US officials on edge”. CNN. https://www.cnn.com/2026/07/31/politics/sweeping-cyberattack-us-water-systems
3. Brumfiel, Geoff. (2026, July 31). “Google pauses AI satellite images, after fears of deepfakes in the sky”. NPR. https://www.npr.org/2026/07/31/nx-s1-5914652/google-adds-ai-to-satellite-images-raising-fears-of-deepfakes-in-the-sky?utm_medium=social&utm_campaign=npr&utm_term=nprnews&utm_source=threads.net.
4. Funk, Dennis. Svirnovskiy, Rennie. Powers, Martine. (2026, January 29). “The quest to ‘destructively scan’ all the world’s books”. The Washington Post. https://www.washingtonpost.com/podcasts/post-reports/the-quest-to-destructively-scan-all-the-worlds-books/
5. Oremus, Will. (2025, September 5). “AI firm Anthropic reaches landmark $1.5B copyright deal with book authors”. The Washington Post. https://www.washingtonpost.com/technology/2025/09/05/anthropic-book-authors-copyright-settlement/
6. Bartz, Andrea, Andrea Bartz, Inc., Graeber, Charles. Wallace, Kirk. (2025, September 5). UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF CALIFORNIA SAN FRANCISCO DIVISION. https://storage.courtlistener.com/recap/gov.uscourts.cand.434709/gov.uscourts.cand.434709.362.3.pdf
7. Tixon, Bernadette B. (2026, July 28). “Inside Project Panama, Anthropic’s Secret Effort To Scan and Shred the World’s Books.” International Business Times UK. https://www.ibtimes.co.uk/anthropic-secret-book-scanning-operation-1811155.
8. https://www.courtlistener.com/docket/69058235/554/21/bartz-v-anthropic-pbc/
9. Edwards, Benj. (2025, June 25). “Anthropic Destroyed Millions of Print Books to Build Its AI Models.” Ars Technica. https://arstechnica.com/ai/2025/06/anthropic-destroyed-millions-of-print-books-to-build-its-ai-models/.
10. Hansen, Dave. “Anthropic Wins on Fair Use for Training Its LLMs; Loses on Building a ‘Central Library’ of Pirated Books.” Authors Alliance, June 24, 2025. https://www.authorsalliance.org/2025/06/24/anthropic-wins-on-fair-use-for-training-its-llms-loses-on-building-a-central-library-of-pirated-books/.