• Grand Theft Auto 6 can take up to 80 hours to beat
• Rob Nelson, the head of development, says it took him 80 hours, including the main story and some “optional goals with narrative ramifications,” to complete the game
• This makes GTA 6 Rockstar’s longest game yet

I hope Grand Theft Auto 6 fans have booked enough time off work, because the game will take up to 80 hours to beat, and that’s just the main story.

In a new interview with The New York Times, Rob Nelson, the head of development and co-studio head at Rockstar North, revealed that his playthrough took 80 hours to finish, including the main story and some “optional goals with narrative ramifications.”

A hot potato: X says that it has just identified a bot farm of approximately 200,000 accounts created by Chinese operators. What’s getting a lot of attention is that 200 of these accounts were apparently dedicated to posting anti-AI data center content designed to influence public perception. The country has tried this before with little success – most people already hate the facilities, anyway.

X’s Global Government Affairs writes that the 200 accounts were posting “in a manner that could manipulate a legitimate debate about American AI and energy policy.”

The posts contained “claims” that AI data centers are driving up household electricity prices and straining the grid. Other posts featured AI-generated cartoons that depicted data-center operators enriching themselves at the public’s expense.

– Global Government Affairs (@GlobalAffairs) August 28, 2026

At least two of the example images appeared in a similar OpenAI report from June. It claimed that China-based ChatGPT users asked the chatbot for ways to influence Americans’ opinions of data centers in the US. The resulting posts also claimed that AI data center buildouts for AI were increasing electricity prices for average families.

When discussing rising electricity prices, the OpenAI report said the accounts “claimed” this was the case. But it really is happening, as noted by Bloomberg, which found that prices have risen by up to 267% in areas near data centers.

Goldman Sachs came to a similar conclusion. It wrote that electricity prices have already increased by 6.9% year-over-year as of February 2026, and they’ll continue to increase as more AI data centers are built.

X didn’t say what the remaining 198,800 accounts were posting. OpenAI’s June report said the other main content criticized US tariffs and technology policies, so one could assume the same thing was happening here.

The pushback against data centers in the US has been growing in both size and intensity. It’s led to claims that the opposition hurts the economy and national security. Fox News compared it to protesting against tank and plane factories during World War II. Even President Donald Trump said those opposing the construction of new data centers are “making a mistake.”

Ultimately, it feels like China is wasting its time trying to convince people of something they already believe. The facilities consume enormous amounts of electricity and can strain power grids that were never designed for such concentrated demand. One planned Michigan campus is expected to draw 1.4 gigawatts – roughly the output of a nuclear power plant – while residents have also raised concerns about traffic and the loss of agricultural land.

Water use is another major concern. Virginia data centers consumed at least 1.85 billion gallons in 2023, up from 1.13 billion gallons in 2019. In Georgia, one development used 29 million gallons of water before receiving a bill while nearby residents complained about low water pressure.

The expansion can also bring constant noise and more pollution, especially as operators turn to diesel generators and gas turbines when the grid cannot keep up. China may be exploiting the controversy, but it certainly didn’t invent it.

For years, the United States has treated its reading crisis as an early grades problem. This narrow focus misses the bigger challenge: the “comprehension wall” that many students hit in middle school. 

As students move into middle school, they no longer focus on learning to read. They must understand what they are reading. Right now, they struggle to do so: Only 30% of eighth graders were proficient in reading in 2024, the lowest level in 30 years.

I witnessed this problem firsthand as a principal and school network leader. We invested significant time in the early grades teaching and reinforcing phonics, phonemic awareness and fluency. The work paid off, as students made strong gains through fourth grade. Then, in fifth grade, performance dropped as texts became more complex and writing demands increased.

That is when many students hit the “comprehension wall.” They can decode the words on the page, but meaning breaks down when a passage uses complex sentences, sophisticated vocabulary, old-fashioned language, hidden meaning or unfamiliar subject matter.

The consequences are hard to miss. Students who cannot deeply engage with text can fall behind not only in English language arts, but science, social studies, math and every subject that depends on reading to learn. Their confidence can erode, with many choosing to hide their confusion rather than ask for help.

The diverse needs that exist in almost every middle school classroom make the challenge harder. Some students struggle to decode, some can read the words but lack the vocabulary or background knowledge to understand the meaning, and some read far above grade level and need to be challenged. 

The problem is not a lack of educator commitment. It is a lack of scalable support.

Opportunities and Challenges

I am generally skeptical of the artificial intelligence (AI) hype in education, especially of suggestions that technology can replace teachers, strong curriculum or the critical relationships that often power learning. But middle school reading is one area where carefully designed AI-supported tools can help if they are built to strengthen instruction rather than shortcut it.

The opportunity lies in custom-built literacy tools, based on the science of reading, that use AI to diagnose student needs precisely, provide timely feedback and help teachers differentiate support without lowering expectations.

The opportunity is not to hand students a generic chatbot and hope comprehension improves. The Brookings Institution found that AI can either enrich or diminish student learning depending on how it is used, and recommended schools use tools that “teach, not tell.” 

Schools can use AI to create better support for the reading work educators already know students need: a higher volume of reading, more precise diagnosis, timely feedback, opportunities to revise their thinking and ways to access grade-level texts when the work is difficult.

One of the biggest challenges for teachers is reading assessment. In middle school, assessment often focuses on whether students have mastered a standard. But reading comprehension is not a simple sequence of discrete skills. A student may understand the main idea in one unit and struggle with that same standard in the next because the texts are more complex, the vocabulary is less familiar or the background knowledge demands are higher.

Educators need a more nuanced picture of what students need. Instead of relying only on stand-alone assessments, well-designed AI tools can analyze students’ everyday work, identify text-specific and longitudinal patterns and give teachers clearer visibility into where comprehension is breaking down.

AI can also help with differentiation. For example, students could read a complex nonfiction article in class and work in pairs to unpack its meaning; but if they struggle, AI asks a just-right follow-up question. They are redirected to a specific sentence or paragraph to reread. They can access a vocabulary definition or brief background knowledge. They have a chance to revise after feedback that pushes them back into the text rather than away from it.

One teacher often cannot provide this level of individual support in real time to 26 students with different needs. There can be a greater impact on struggling readers, who are too often removed from rigorous learning and complex texts altogether, and are given watered-down materials instead of better access points. Over time, the achievement gap becomes not just a reading gap, but also a knowledge gap.

Early results are promising. In one large urban district in Texas and another very diverse district in Tennessee, middle schools using an AI-supported literacy model saw significantly stronger year-over-year growth on state reading assessments than comparable schools that did not use the model.

These examples suggest AI-supported literacy intervention can help educators provide more responsive support at-scale when aligned with strong instruction, grade-level content, teacher support and consistent implementation.

If AI becomes a shortcut, novelty, or substitute for high-quality instruction, it will fail students. If it is used mainly to help students complete assignments faster, it will weaken the habits students need to build: reading closely, asking questions, explaining their thinking and revising their ideas.

Ask Before You Act

The best way to improve student reading achievement is still to read a lot, and we should look for tools that support a higher volume of reading with increasingly complex texts.

In the last decade, schools have taken a crucial step and replaced incoherent, homegrown materials with vetted curricula that systematically build knowledge and vocabulary, expose students to complex grade-level texts and spiral essential reading habits.

Before adopting AI-supported reading tools, educators and district leaders should ask hard questions, such as:

• Does the tool reinforce or detract from high-quality curricular materials? Does the tool keep students engaged with grade-level text or does it lower the bar? 

• Does it require students to return to the text, explain their thinking and revise their responses? 

• Does it give teachers visibility into where students are struggling? 

• Does it support discussion, writing and teacher-led instruction or does it isolate students on screens? Does the tool lead to more students doing more reading, writing and thinking? 

• Does it help students build reading stamina and a love of reading?  

These questions matter because the goal is not to make reading easier by removing the cognitive work. The goal is to make rigorous reading more accessible so more students can participate in that work.

We cannot continue treating middle school reading challenges as an inevitable byproduct of earlier gaps. Nor can we assume that students who reached adolescence without strong literacy foundations are beyond meaningful intervention.

For middle school students who have spent too long being told that reading is something they should have already mastered, that support could make all the difference.

Flash memory is fast, compact and everywhere, but its biggest drawbacks have never completely disappeared.

Why flash memory became so popular

monte_a/Shutterstock

The greatest drawback is that flash wears out

Screenshot by Husain Parvez (Samsung)

What could replace flash memory?

Alberto Garcia Guillen/Shutterstock

Henry van Dyck of Veritasium wanted a number, not a slogan. Classroom posters and superhero movies have spent years saying silk spider webbing beats steel. In a video he co-wrote and co-hosted with Derek Muller, he took that claim into Professor Todd Blackledge’s spider lab at the University of Akron and put live threads on a tensile tester, without harming the animals that made them.

Garden variety dragline silk, the cable a spider uses to keep from falling, and the framework of a web are all very impressive. When you put a 600 megapascal number on a rope, you get a sense of how strong it actually is. A silk thread with a cross section the size of a little finger can support around 60,000 newtons, which is comparable to the weight of an adult African elephant. Van Dyck was able to replicate that in the lab, but emphasized that this was simply an average occurrence.

Insta360 X6 Essentials Bundle – 8K 360 Camera with 1/1.1” Sensors

• DUAL 1/1.1″ SENSORS & TRIPLE AI CHIP SYSTEM: Custom-built dual 1/1.1″ square sensors deliver unprecedented 360 image quality. Combined with a triple…
• FLAGSHIP 8K50FPS 360 VIDEO: Hit record on our 360 action camera and capture your world in stunning native 8K50fps video. Every view, every highlight…
• IN-HOUSE AI MODEL FOR EFFORTLESS 360 EDITING: Insta360’s own AI model, built specifically for 360 editing. New AI Director edits your footage…

Darwin’s bark spiders in the wild, deep in Madagascar, weave some magnificent webs across rivers up to 25 meters wide to snare unsuspecting insects flying over the water. Their dragline silk has been measured at an incredible 1,600 megapascals, more than twice the strength of the average Akron sample, making it the toughest biological fiber we’ve ever measured, even beating out the strongest experimental ultra-high-strength steels that come in at around 3,000 megapascals, but as soon as two wires of equal thickness are matched, those steels break, whereas the spider’s silk holds up just fine.

The density of the materials tells a quite different story, as steel is around six times heavier than silk for the same volume. When the weights of two materials with the same mass and length are compared, the cross section of silk rope is six times bigger than that of steel. Six times the area, at nearly a third of the breaking tension, explains why silk can bear approximately twice the force before collapsing. Engineers refer to this as specific strength, and ordinary spider silk is already competing with fancy steel on this front.

Toughness is a whole other thing, and silk wins hands down. The area under a stress-strain curve indicates a fiber’s toughness, or how much energy it can withstand before breaking. Kevlar is reasonably effective at stopping bullets, with a maximum energy density of around 50 megajoules per cubic meter. The experimental ultra-high-strength steel weighs in at 170. Van Dyck’s garden silk measured around 205 megajoules, whereas Darwin’s bark spider silk measured up to 520 megajoules, which is roughly ten times more than Kevlar and three times more than experimental steel. Which is fantastic since silk is all about stretching. Kevlar, on the other hand, is stiff like a board. A stiff fiber will rise up the graph quickly and break early, whereas a flexible fiber will simply continue to take the strain as it lengthens, which is where all the extra energy comes from.

Spiders produce this extraordinarily strong fiber from a liquid protein sludge held in silk glands. As the liquid flows out of the spinnerets, long molecules known as spidroins begin to fold themselves into hard crystalline patches for strength and looser, flexible regions for… well, give. There are no explosives or high-pressure mills, only room temperature, a bug diet, and a healthy dose of natural chemistry. Each visible strand actually a bundle of even smaller strands, literally nanostrands, running in parallel, which is why a thread thinner than a hair may still function as a cable.

Farming animals to make spider silk isn’t a feasible solution. Spiders cannot simply be cultivated like cattle. They require plenty of space, and if kept in a tight environment, they will eat each other. Each spider produces only a few hundred feet of useful thread. Those high-end showpieces, such as 18th-century stockings or the iconic cape constructed from a million spiders in Madagascar, were always one-of-a-kind and never mass-produced. Pure natural silk costs millions of dollars per kilogram.

Kraig Biocraft Laboratories, on the other hand, decided to eliminate the spiders entirely in favor of genetically engineering silkworms to generate spider silk. They take silkworms, a super common moth that humans have farmed for thousands of years, and just add a tiny portion of spider DNA to the mix. Van Dyck spent the day at their lab, watching transgenic cocoons come off the line. Last year, they produced almost half a ton of hybrid silk. The good news is that mechanical testing have showed it performs at roughly 60% the level of real spider silk, which is excellent enough to pique the interest of those who manufacture super-tough ballistic panels and medical mesh.

Then Van Dyck decided to pull off a publicity stunt of sorts. He took a bundle of synthetic silk, including ten threads all lined up, to the local climbing gym and requested Muller to try it. He asked Muller to hang from it and swing for a moment. The line held up nicely, although the skin was damaged. The fiber is so thin that the contact patch on a hand becomes a cutting edge. Muller’s palm paid for the demonstration. In materials language, the silk was no longer the weak link, the person holding it was.
[Source]

Hasbro, one of the world’s largest toy and game companies, has disclosed that attackers have accessed the personal and financial information of an undisclosed number of employees.

Founded in 1923, Hasbro is a publicly traded American multinational entertainment conglomerate on the NASDAQ and owns many brands, including Monopoly, Clue, Nerf, Transformers, Play-Doh, Peppa Pig, Scrabble, Magic: The Gathering, Dungeons & Dragons, and many others.

The company has filed data breach notification letters with the Massachusetts Attorney General’s Office, but didn’t disclose the total number of affected individuals or when the incident was detected.

“The information involved varied by individual but may have included your name and one or more additional personal information elements such as email, address, phone number, national ID number, or financial information,” Hasbro said.

“Hasbro implemented containment and remediation measures, including disabling the compromised employee account, terminating unauthorized access, and deploying additional safeguards designed to help prevent a similar incident from occurring in the future.”

While the breach notification letters don’t detail the total number of affected people or the nature of the information exposed in the incident, the breach affected the Social Security numbers, financial account information, credit/debit card numbers, and driver’s license information of 436 Hasbro employees in Massachusetts, according to the Massachusetts Attorney General’s Office 2026 Data Breach Notification Report.

​A Hasbro spokesperson was not immediately available for comment when BleepingComputer reached out to ask whether any customers were also affected by this breach and whether the attackers sent a ransom demand.

In early April, Hasbro also disclosed a cyberattack that hit its systems on March 28 and forced the company to take some of them offline while working to restore them.

In a filing with the U.S. Securities and Exchange Commission (SEC) at the time, Hasbro warned investors of “some delays,” and said that interim measures taken for business continuity “may continue for several weeks before the situation is fully resolved.”

According to financial reports filed by Hasbro since then, the company has lost approximately $25 million in revenue because of the cyberattack.

Hasbro didn’t link the March incident with the data breach disclosed in notification letters filed with the Massachusetts attorney general’s office this week.

Overall prevention scores can hide what happens after initial access. Once attackers are using valid credentials, prevention drops sharply.

The Blue Report 2026 measures defenses technique by technique across 338 million simulations run in customer production environments.

Get the report

Authorities in Australia said Wednesday that they arrested two men accused of participating in cybercrimes for TeamPCP, a prolific group of hackers that, over nine months, has carried out a relentless series of supply chain attacks that infected more than 1,000 organizations worldwide.

In a statement, the Australian Federal Police said the two men were arrested and charged with 14 offenses. The statement said the men were members of TeamPCP, which by the authorities’ count, compromised more than 1,000 organizations worldwide. The statement didn’t identify the men, except to say they lived in the Western Australian towns of Cottesloe and Mandurah. KrebsOnSecurity, citing a lengthy investigation, provided what it reports to be both defendants’ names, along with an extensive background of their lives and the mistakes that led to their downfall.

The hacks that keep on hacking

TeamPCP has vexed law enforcement officials and security personnel around the world since it emerged in December. The group is best known for a sustained series of supply chain attacks that laced open source software with malware that self-propagated from one package to another. The viral infections worked by targeting organizations’ CI/CD pipelines, which are used to rapidly develop, update, and deploy software.

Once a package or tool was compromised, Shai-Hulud, as the worm was dubbed, attached itself to future package updates. When developers downloaded the compromised packages and ran them through their own CI/CD platforms, their software was also compromised.

Blacknut cloud gaming is coming to TiVo OS, granting users access to more than 1,000 games directly through compatible smart TVs without needing a dedicated console or downloads.

The cloud gaming service has partnered with TiVo to bring its catalogue to Powered by TiVo devices, expanding its reach to brands that includes Sharp, Metz and Thomson.

The service is designed to sit alongside the movies, TV shows, sports and streaming apps already available through TiVo OS.

Identity is the fundamental currency in any technology transaction so we need to treat nonhuman assets with the same care as we apply to people.

We’re outnumbered! Nonhuman identities (NHIs) outnumber human identities in the enterprise by a ratio of up to 50:1. This Non-Human Identity Management Group statistic from 2025 is likely already outdated, and in a year from now the ratio will be exponentially higher as rapid growth trends like AI/Machine Learning, the Internet of Things, digital assistants and the burgeoning API Economy continue to accelerate automation.