Wiz and Google announced Project Atlas, a multi-model solution that outperforms Mythos on a key vulnerability discovery benchmark.
getty
Wiz’s new AI system just beat Mythos and GPT-5.5 Cyber on a key vulnerability discovery benchmark. The cloud security provider, acquired by Google for $32 billion in March 2026, today announced Project Atlas, an AI-driven system for vulnerability research that scored 90.9% on CyberGym, beating Mythos at 83% and OpenAI’s GPT-5.5 Cyber at 85%.
… %!s()
CyberGym is a benchmark that tests the ability of AI agents to discover vulnerabilities, so the fact that Project Atlas outperformed Mythos suggests it could be a more effective defensive tool than Anthropic’s unreleased frontier model.
According to Nir Ohfeld, head of vulnerability research at Wiz, told me in a video interview that one of the key differentiators of Atlas is that “unlike Mythos or Daybreak, Atlas is not a model. It’s a system that uses multiple models.” The tool uses a collaborative network of specialized, task-oriented AI agents, which select the best model for a given task.
The release comes less than two weeks after Hugging Face was breached by an autonomous agent, an incident which OpenAI later claimed occurred when GPT 5.6 Sol and an unknown prerelease model escaped a sandbox environment and hacked the open-source model provider.
Not One Model
When Anthropic announced Mythos back in April, the AI lab claimed the model had already found thousands of high-severity vulnerabilities in every major operating system and web browser. It also launched Project Glasswing as an “urgent attempt” to put these capabilities to use for defensive purposes, allowing third-party companies including AWS, Apple, Microsoft, Cisco and CrowdStrike to experiment with using the model.
MORE FOR YOU
Wiz has also been experimenting with AI to mitigate vulnerabilities in critical software. According to information provided by the company, Wiz first began working on Project Atlas after the company’s ZeroDay Cloud event, held at BlackHat Europe in December 2025. The team shared that during the event, they discovered a record-breaking number of CVEs affecting critical infrastructure, with many of them found using AI.
After discovering how much vulnerability research was shifting toward AI-assisted systems, the company began building Atlas, before joining Google to collaborate on what it refers to as “cutting-edge AI models for cyber defenders”. Since joining Google, Wiz has worked with multiple teams across the company, including DeepMind, to improve the tool.
Ohfeld notes that Wiz is using ChatGPT and Opus as part of the solution, but notes that Atlas scores above both in benchmarks because “we are using each model for its specific strengths." Not only has the approach performed well on CyberGym, but Wiz also claims that Atlas has found more than 200 previously unknown vulnerabilities in open-source projects, with fewer hallucinations and false positives than other alternatives.
He added that earlier this year, an early version of Atlas found “the most severe vulnerability ever found in GitHub.com,” helping to patch a vulnerability that would have allowed an attacker to run commands on GitHub servers and access customer repositories. This resulted in a bounty payout of $100,000, the largest bounty paid out to a researcher.
Inside Atlas
While powerful offensive models like Mythos and GPT-5.5 Cyber haven’t yet been widely released, with access gated behind Anthropic’s Project Glasswing and OpenAI’s Trusted Access for Cyber, these capabilities can’t be kept off the market forever. After all, equally strong models will come to market, as GLM-5.2 has shown.
As a result, defenders must prepare for a world where vulnerabilities are exploited at machine speed. Atlas represents an alternative approach to securing software against these fast-moving threats. “
Project Atlas is a meaningful demonstration of how quickly AI-driven vulnerability research is moving from experimental model capabilities toward operational security systems," Katie Norton, senior research manager of DevSecOps and software supply chain security at market intelligence provider IDC, told me via email, noting that frontier AI models can already reason about vulnerabilities.
However, Norton warns that “discovery is outpacing the industry’s ability to coordinate remediation and disclosure in the open source ecosystem,” adding that initiatives like Linux Foundation’s Akrites and the White House’s Gold Eagle reflect growing recognition that AI-generated vulnerability findings cannot simply be delivered by overburdened project maintainers.
“The industry will need shared mechanisms to validate and deduplicate findings, coordinate fixes confidentially, prioritize the vulnerabilities that present the greatest risk, and help patches reach downstream users before disclosure creates additional exposure,” Norton said.
Vulnerability Discovery Is Changing
Over the past year, Google has developed a number of vulnerability discovery tools, including CodeMender, a tool initially released as a research project in October 2025, which can automatically find and fix code vulnerabilities, and also Gemini 3.5 Flash Cyber, a lightweight model designed to find and patch vulnerabilities, announced in July.
Atlas appears to be Wiz and Google’s answer to a threat landscape that’s increasingly moving at machine speed. Ami Luttwak, cofounder and CTO of Wiz, told me in a video call that Atlas outperforming Mythos “shows you that for vulnerability scanning, it’s not just about the model, it’s also about the system and about how you use different models together.”
Luttwak adds that “there is urgent work that defenders need to do" following the Hugging Face Breach.“The same system two years ago would be considered secure, but the same system right now is not secure if you’re not running the latest AI model on your applications from the outside and on your code."
With 48,000 new CVEs discovered in 2025, defenders need to be prepared for threat actors to be able to exploit vulnerabilities at scale with AI. Tools like Mythos and Atlas could potentially provide an answer to this increase in scale if defenders can discover how to use them to patch exposures more efficiently. %!s()
Craig Crisler is CEO of SupportNinja and has 20+ years operational excellence and modern people management experience.
getty
Companies have never lacked customer data.
Every customer interaction leaves behind clues: where people get stuck, what frustrates them, what causes them to leave and what keeps them coming back.
… %!s()
The challenge? Surfacing the signals that actually matter.
That’s where AI changes the conversation.
AI can analyze enormous volumes of customer interactions and identify patterns, friction points and operational issues at a scale that simply wasn’t possible before. AI gives companies the visibility to understand customers more deeply and improve the experiences they create.
AI Makes Customer Friction Visible
Consumers already expect personalization in nearly every interaction.
Streaming platforms anticipate preferences. Retailers recommend products based on purchase history. Airlines proactively rebook disrupted travel. Customers increasingly assume companies should understand who they are and what they need before they ask.
With AI, companies can now analyze behavior patterns with a level of granularity that was previously impossible and rapidly identify specific friction points across the customer journey.
Using AI to better understand customer intent, predict needs and eliminate unnecessary effort gives companies the visibility to continuously improve the customer experience.
Customers can already feel the difference between automation that reduces friction and automation that creates it.
AI Doesn’t Fix Broken Experiences
Despite billions invested in CX initiatives over the past decade, customer satisfaction remains stagnant.
A big reason why? Too many companies still approach CX as a cost center instead of a business strategy.
Since technology scales what already exists, AI risks amplifying that problem.
Problems arise when companies rush to deploy AI without redesigning the underlying customer experience.
Low-quality chatbots, endless support loops, rigid automation and inaccessible human support may improve operational efficiency, but they do so at the expense of customer trust.
Automating a broken workflow just helps companies to deliver a bad experience faster and at scale.
That’s why visibility matters.
Before companies automate a process, they need to understand where customers experience friction, why it exists and how removing it will improve the overall experience.
AI Is Most Valuable As An Intelligence Layer
The biggest opportunity in AI is visibility.
AI can analyze massive amounts of customer interaction data and surface operational insights that previously required extensive manual analysis. Leaders can identify where customers abandon journeys, where support friction emerges, which workflows create repeat issues and where the experience companies promise doesn’t match the one customers actually have. They can then fix these problems.
That visibility changes the conversation.
Instead of reacting to customer dissatisfaction after the fact, companies can proactively identify systemic problems before they scale. AI becomes less of a front-line replacement tool and more of an operational intelligence layer that continuously improves decision-making.
One of the most practical ways to begin is by asking AI to analyze customer interactions, identify friction throughout the customer journey and recommend the highest-impact opportunities for improvement.
AI Makes Human-Centered CX More Valuable
As AI becomes more common, the experiences that stand out will feel more human, not less.
Customers might not know whether an interaction was AI-enabled. But they do know when it feels helpful. They know when it reflects the company’s brand. And they know when it feels generic.
That’s why human-centered CX becomes even more important in an AI-driven world.
Every AI-powered interaction, from chatbots to knowledge bases, should reflect the same voice, values and experience customers expect from every other touchpoint.
Technology creates consistency. People create connection. The strongest customer experiences require both.
Visibility Changes Everything
As AI becomes more capable, the companies that benefit most won’t necessarily be the ones automating the most work. They’ll be the ones gaining the clearest understanding of their customers.
Visibility changes how companies identify friction. It changes how they prioritize improvements. And ultimately, it changes how they make decisions.
AI gives companies an unprecedented opportunity to understand customers more deeply than ever before. The leaders who use that visibility to remove friction, strengthen relationships and continuously improve the customer experience will be the ones best positioned for the future.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
In the projector market, Xgimi has been making a name for itself with compelling lifestyle projectors that offer competitive projection alongside stylish designs that better blend in with home decor than the ugly plastic boxes coming from many other brands. With the earlier Titan, Xgimi tested the prosumer waters. Now, the Titan Noir family finds a middle ground, offering something for enthusiasts who want superb projection (and more screen size than a standard 4K TV) in a living room-friendly package. The Titan Noir family includes the Titan Noir ($3,999), Noir Pro ($4,999) and Noir Max ($5,999) with the key differences between each model being their brightness and contrast ratio. Xgimi sent over the middle sibling for testing, and whether for watching movies or gaming on the big screen, it downrights rocks.
…
Xgimi Titan Noir Pro – Design and Features
The Xgimi Titan Noir Pro is a looker. Xgimi has shown a knack for fairly elegant-looking lifestyle projectors that step away from the dated designs that look more at home in an office. The Titan Noir Pro follows that trend, combining smooth metal and stylish grilles. It looks less at home on a coffee table than some of Xgimi’s other models, but it’s still somewhat tame for a high-end home theater projector.
It’s also shockingly small and light for a projector delivering as much as it does. I’ve tested models like the Epson LS12000 and Sony Bravia Projector 7, and those things were big units that took up a huge swathe of my mantle and took some care to heft up there. By contrast, the Titan Noir Pro is only 9.1 inches tall, 11.3 inches deep, and 14.9 inches wide, and some of that size comes from the metal legs it stands on. It’s also just 16.75 pounds, and that frankly feels almost too light for a projector aiming to provide so much brightness.
Aside from the legs that the Titan Noir Pro stands on, it’s a tame design. It’s really just a box with various vents for air intake and exhaust, a couple of speakers directly in the front. The lens sits off to one side at the front flanked by a few extra sensors for handling things like autofocus, auto keystone, and an eye-protection tool that dims the output when someone is in front of the laser projection. That last feature is useful, and this model has a special version that tries to only dim the light around where it detects a person’s head, though it doesn’t do a great job compensating when lens shift has been used on the projection.
One major advantage for the Titan Noir Pro is its optical system. The optics of the projector offer zoom and lens shift to a considerable degree (throw ratios range from 0.98:1 to 2.0:1, and the projection can shift 50% horizontally or 130% vertically in either direction). This makes it possible to set the projector in one spot and shrink and move the projection to where it fits quite easily without sacrificing a single pixel. By contrast, many projectors advertise convenient automatic screen alignment but do so digitally, cropping the projection to fit and lowering the resolution you see as a result.
On the back of the Titan Noir Pro are most of the ports you’ll need. There are three HDMI ports. One of them offers eARC for audio, though there’s also a 3.5mm and an optical audio output port. There are two USB-A ports for local media or powering streaming sticks. The projector includes a LAN port and Wi-Fi connectivity, though you might not get much use out of either because the projector lacks Google TV or the like.
Xgimi does include a large styrofoam carrying case that’s helpful for protecting the projector on the go and keeping its remote and power brick all together. That power brick is unfortunately very large and heavy. It would have been convenient to have the power supply integrated into the projector, but I’ve begun to suspect some brands keep them external on premium products like this so they’re easier to replace in the event of failure.
The remote included is also helpful. It provides simple controls for easy navigation of the settings and offers backlighting on a few of the primary buttons. The remote also includes four customizable shortcut buttons that let you quickly access specific settings you’ll use often. The volume buttons are easy enough to feel out in the dark of a home theater, but both the volume and shortcut controls don’t get any backlighting.
Xgimi Titan Noir Pro – Software
Many of Xgimi’s projectors come with a built-in streaming OS like Android TV or Google TV, but not the Titan Noir Pro. It somewhat feels like it has Google TV built-in, because a lot of the menus are similar, but that’s all the operating system handles: settings menus. This means you’ll need to connect video sources (or a USB drive) to get anything out of the projector, but it also makes for a very responsive system that allows for quick changes to settings and faster bootup.
It might have been nice to see a fuller operating system baked in, but in a time when LG monitors are delivering adware and operating systems are plugging large banner ads on startup, I’m starting to think that maybe the better default is actually to not include one. Xgimi could have at least provide a landing screen, perhaps with previews of each HDMI input source, which can be helpful when you – like me – never remember which device is connected to which HDMI port.
One setting I wish the projector supported is the option to skip HDMI 1 as an input source. That port handles eARC audio output, so there’s a good chance it will be used for audio output exclusively in many setups, but the projector still lets you select it was an input source even when it’s handling eARC (sure, some eARC audio devices can also pass through video, but I don’t think most are used in that way), Since HDMI 1 is the first input source the projector boots, too, it won’t be useful at all for folks who just have that connected to their sound system.
Xgimi Titan Noir Pro – Gaming and Performance
The Xgimi Titan Noir Pro is simply brilliant. I’ve been using and testing projectors for a long time, and the Titan Noir Pro may be the most well-rounded and capable model I’ve put my hands on.
To start, its adjustable optics make a perfectly sharp picture easy to get without having to fuss much with where the projector is positioned. Whether low on a coffee table, mid-way up the room on a desk, or perched high on a mantle, I could easily get the projector images squared away on my wall or projection screen using the zoom and lens shift controls. From there, autofocus can sharpen the image quickly and effectively.
Not needing keystone adjustments is huge, as these not only cut into the usable image but also introduce latency, and for gaming, that’s just unacceptable.
That said, the out-of-the-box defaults aren’t exactly the best. I’m always surprised to see high-end projectors like this come running heavy motion interpolation by default, and the Titan Noir Pro is no exception. It introduces unsightly artifacts and awkward motion in a lot of cases. If you’re gaming, the projector will likely automatically switch to game mode that disables this, but with a streaming stick connected, it doesn't.
The projection itself is simply stunning, though. The Titan Noir Pro is fantastically bright. I played through several 007 First Light challenges, including some that took place at night, even while light came in from outside my living room. I was content watching movies and TV without even turning off all the lights, too.
Of course, the Titan Noir Pro looks even better in a dark space. It’s normally too bright to actually use in a dark room (think lights out and blackout curtains in place). But it features a dual-iris system that can clamp down to reduce the overall light output while ramping up contrast. The result is a picture that still can dazzle with highlights while showing off a deep black floor. On a lot of projectors I’ve tested, I can see black letterboxes no matter what’s on-screen, but with the Titan Noir Pro, I could almost never see the letterboxes.
All the brightness and contrast comes by way of an RGB laser light source that simply dazzles when it comes to color output. While a projector may never match an OLED display in terms of contrast, RGB laser light sources offer simply staggering color gamut. The colorful scenery, costumes, and bending effects in the live-action Avatar: The Last Airbender show are stunning to see on the Titan Noir Pro. With support for HDR10+ and Dolby Vision, the Titan Noir Pro has you covered for high-end formats. It’s too bad it’s a little early for Dolby Vision 2 support.
Those visual chops come along for gaming, too. I played through hours of 007 First Light on the Titan Noir Pro, and it kept up brilliantly. The story beats were all rendered with cinematic flair, the projector making it seem all that much more like a movie. Latency was negligible while running at 4K/60Hz, but the projector can also drop down to 1080p/240Hz and enable VRR (AMD FreeSync Premium). In that setting, I smashed through the fast-paced action setpieces of the game with imperceptible input lag, making it easy to dash for cover and snap to enemy targets. I normally game on fast OLED monitors with keyboard and mouse, and gaming on the Titan Noir Pro didn’t feel like it held me back one bit. The Xgimi Horizon 20 Max has some similar capabilities at a lower price and will likely work well if you’re mostly after brightness, but the Titan Noir Pro’s black levels are on another level.
Alongside the great visual performance, the Titan Noir Pro is surprisingly quiet. Though it has a good deal of power to contend with, I never heard its cooling system even from a few feet away.
Impressive though the visuals may be, audio from the projector is nothing to get excited about. The speakers can put out a bit of volume, but they don’t create a sense of space, don’t have deep sub-bass, and are too focused in the mids, creating a harsh sound at high volume. They can do in a pinch, but you’ll absolutely want to pair the projector with speakers to get the most out of it.
Xgimi Titan Noir Pro – Testing
Analyzing a projector is a little more difficult than a display, since you can’t simply set up a colorimeter on the lens, and anywhere else the colorimeter will either block the light it’s meant to measure or capture ambient lighting as well. That doesn’t make it impossible to evaluate, but less precise. Still, with my colorimeter set up a foot in front of a 1.1-gain white projector screen, it measured the Titan Noir Pro at 93% coverage of the DCI-P3 color space and 85% of the BT.2020 color space. Having seen my fair share of ultra-colorful displays, I’d say those measurements are on the low side of what I’m actually seeing from the Titan Noir Pro.
Brightness is a little more direct to measure. I displayed a full white screen and averaged the brightness across nine points to find the Titan Noir Pro putting out 4,375 lumens. That was with the Standard picture setting and iris fully open, which thankfully doesn’t lead to a picture that’s overly blue or green as some projectors in their highest brightness modes do.
Contrast is more difficult to measure but not impossible. Switching to a test pattern for measuring contrast with two bright white squares in opposite corners and black everywhere else, I measured an average 4,870 lux for the white areas and just 5 lux for the black space, coming out to a 974:1 contrast ratio. That’s already incredible for a projector, but the Titan Noir Pro still has some moves to bust out.
With its iris shut down as far as possible, its brightness goes down considerably, but its contrast jumps. The same test produced an average 1,238 lux in the corners and 0.77 lux in the center for a 1,608:1 contrast ratio. Here is where we get to the challenge for epic projectors like this: you need to optimize the viewing area as well. A lot of light will bounce off your wall or screen, reflect throughout the room, and land right back on the screen, elevating the black levels. By blocking out most of the projection with a black velvet fabric that absorbed most of the light but had a center hole to allow some of the projected light through, I was able to measure the black center point at an even lower 0.19 lux with all settings otherwise the same as above. This would put the contrast at an exceptional 6,518:1 in a thoroughly light-controlled room.
Mark Knapp is a contributing freelancer for IGN covering everything electronics and gaming hardware. He has over 10 years of experience in the tech industry with bylines at PCMag, Reviewed, CNET, and more. Find Mark on Twitter @Techn0Mark or BlueSky at @Techn0Mark.
Verdict
The Xgimi Titan Noir Pro is phenomenal. Its RGB laser projection is a sight to behold, and it has the versatile optics to ensure that projection is perfected for a huge range of setups, letting it work around you rather than forcing you to work around it. It provides an exceptional way to enjoy both movies and games, especially thanks to its support for an ultra-responsive 1080p/240Hz mode and VRR. It delivers brightness that can hold up in bright rooms and black levels that call for special room treatment to make the most of. Even at its full $4,999 price, it’s an amazing option that’s competing with projectors far more expensive. If it ever goes on sale near its early-bird Kickstarter price, I’d have a hard time recommending anything else.
You're currently following this author!
Want to unfollow? Unsubscribe via the link in your email.
Warren Buffett (left) and Greg Abel, who succeeded him as Berkshire Hathaway CEO this year.
Nati Harnik/AP, FileBrendan McDermid/REUTERS
…
Warren Buffett's successor, Greg Abel, just completed his first acquisition as Berkshire Hathaway's CEO. It's the latest sign that he's ready and willing to build on what Buffett built.
In a Friday press release, Berkshire said it has officially taken over Taylor Morrison for $8.5 billion in cash, a deal it first announced at the end of May.
Taylor Morrison is one of the nation's biggest land developers and homebuilders; it made about $1 billion in pre-tax profits on around $8 billion of revenue last year, its latest annual report shows.
The homebuilder has now joined Berkshire's stable of housing-related subsidiaries, which includes its real estate brokerage network, Berkshire Hathaway HomeServices, and its building-products businesses, such as Clayton Homes, MiTek, and Acme Brick.
"Berkshire is acquiring a best-in-class national homebuilder, led by an exceptional team and backed by a trusted reputation for customer experience," Abel said in the May press release, echoing Buffett's appreciation for market power, quality management, and a strong brand.
Abel said he expects to "unify our site-built homebuilding operations into a combined platform enabling us to deliver the dream of homeownership to more Americans."
He shared an almost identical vision in Friday's press release, signaling that he plans to marshal Berkshire's homebuilding might to try to help tackle the affordability crisis that has prevented many Americans from becoming homeowners.
The press release laid out that Taylor Morrison's brands — including Esplanade, Yardly, and Taylor Morrison Home Funding — will be combined with Berkshire's site-built homebuilding operations, a set of 15 regional and local homebuilders that make up Clayton Properties Group.
Taylor Morrison's CEO, Sheryl Palmer, similarly said the united operations will have "transformative" scale and reach.
Forging his own path
Abel took over from Buffett on New Year's Day, marking the end of a nearly six-decade run for the legendary investor that saw him transform a failing New England textile mill into a $1 trillion conglomerate. Buffett remains chairman, and both men have said he provides input into every major decision at the company.
Regarding the Taylor Morrison deal, Buffett told CNBC's Becky Quick in May: "Greg did that faster than I could have done it, smoother than I could have done it, and I never talked to the CEO. He has launched."
Buffett's defining struggle for years has been figuring out smart ways to deploy Berkshire's cash pile, which doubled in size to $380 billion in the two years ended March 31. That sum exceeds the market capitalization of many of America's most valuable companies, including General Electric, Coca-Cola, and Procter & Gamble.
The bargain hunter has balked at buying stocks at historic highs, making acquisitions when private-market valuations are lofty and private equity firms present fierce competition, and even buying back Berkshire stock as it climbed to record highs.
The situation may finally be improving. Buffett recently told CNBC that he made the decision to invest in Alphabet last year. During the nine months ended March 31, Berkshire amassed a stake in Google's parent company worth $18.5 billion today.
It bought another $10 billion of stock in a private placement in June, giving it a roughly $28 billion stake today, assuming it hasn't altered the size of its holding in recent months.
That means Berkshire, which rarely makes big changes to its stock portfolio, took Alphabet from scratch to a top-five holding in under a year.
Berkshire also struck a deal to acquire OxyChem in October for nearly $10 billion from Occidental Petroleum, one of the larger holdings in its stock portfolio in recent years.
Alongside his efforts to allocate Berkshire's capital, Abel has also started putting his top lieutenants to work. Michael O'Sullivan, Berkshire's first general counsel, and Charles Chang, who's poised to take over as finance chief from Marc Hamburg next year, briefly served as interim directors of the acquisition vehicle buying Taylor Morrison, along with Hamburg, an SEC filing shows.
It's not totally clear how much influence Buffett has had on Abel's early decisions. But it appears that between making an AI giant one of Berkshire's largest holdings, acquiring a leading homebuilder, and enlisting O'Sullivan and Chang into the merger process, Abel is starting to make his mark.
Read next
Theron Mohamed
You're currently following this author!
Want to unfollow? Unsubscribe via the link in your email.
Theron Mohamed is a London-based correspondent on the Trending team at Business Insider. His coverage spans finance, investing, wealth, markets, and the economy.Theron joined BI in 2019 as a reporter at Markets Insider and rose to the rank of correspondent before moving to the Trending team in 2024. He previously covered tech, media, and telecom stocks for Investors Chronicle magazine and had a brief stint on the Financial Times' Data team. He interned at the Wall Street Journal in New York where he primarily wrote for Heard on the Street.Theron has freelanced for The Independent, The Telegraph, WIRED, and several smaller publications. He holds an undergraduate degree in geography from the London School of Economics, and a master's degree in journalism from Columbia University.Theron often covers Warren Buffett, Michael Burry, Jeremy Grantham and other top-flight investors. He also writes about the world's wealthiest people and shares financial advice from all manner of rich and successful people.Email Theron at [email protected] and follow him on X @theron_mohamed.Expertise
Corporate finance
Stocks and investing
Wealth and philanthropy
Business history
US economy
Warren Buffett and Berkshire Hathaway
Popular articlesAl Pacino says he went from $50 million to broke, joining a long list of stars who've experienced money troublesAn oil tycoon sold his company for $26 billion this year — but died before the deal closedWarren Buffett drinks 5 cans of Coke a day — here's why he switched from Pepsi after nearly 50 yearsMeet the 16 people in the $100 billion club — who are jointly worth more than Amazon or Google'Big Short' investor Michael Burry kept quiet, piled into China tech, and won big with a stock bet in 2024Bill Gates' former assistant is worth $154 billion — and could soon be richer than the Microsoft cofounderHoward Schultz talked about Steve Jobs, trademarking the latte, and Starbucks' problems in a marathon interviewWarren Buffett just made a rare trip to Tokyo. Here's the story of a disastrous sushi dinner that made him swear off Japanese food forever.21 states where recession bells are ringing after unemployment jumpsWarren Buffett is building the Noah's Ark of rainy-day funds. Here's why he's stacked up more than $300 billion.The 'Shark Tank' star Kevin O'Leary warns couples not to combine finances: 'I don't care how in love you are'The Waltons are once again the world's wealthiest family, beating out Gulf royalty and fashion dynasties
%!s()
Why do some dreams feel so vivid that they seem almost real, while others unfold as confusing fragments that are difficult to remember or understand? New research suggests the answer may depend on a combination of personality, sleep quality, mental habits, and major events unfolding in the outside world.
… %!s()
A study led by researchers at the IMT School for Advanced Studies Lucca found that dreams are not simply random collections of images and thoughts. Instead, they appear to be shaped by both individual differences and shared experiences, including major social disruptions such as the COVID-19 pandemic.
Thousands of Dream Reports Reveal Hidden Patterns
The research, published in Communications Psychology, examined more than 3,700 descriptions of dreams and waking experiences from 287 people between the ages of 18 and 70.
For two weeks, participants recorded their dreams and daily experiences. The researchers also collected information about each person's sleeping habits, personality, cognitive abilities, and psychological characteristics.
This broad set of information allowed the team to compare what people experienced while awake with the stories, settings, and emotions that appeared in their dreams.
How AI Helped Scientists Study Dreams
To analyze such a large number of written reports, the researchers used natural language processing (NLP), a form of artificial intelligence that examines patterns, relationships, and meaning in language.
Rather than relying only on researchers to read and interpret every dream individually, the technology allowed the team to measure the semantic structure of the reports. Semantic structure refers to how ideas, words, and concepts are connected within a piece of language.
The results showed that dream content was neither completely random nor inherently chaotic. Instead, dreams reflected a complex mixture of personal characteristics, including a tendency to mind-wander, an interest in dreams, and the quality of a person's sleep. Events occurring in the broader world also influenced what people dreamed about.
Dreams Do Not Simply Replay the Day
When the researchers compared participants' descriptions of waking life with their dream reports, they found that the brain did not appear to reproduce daily experiences exactly as they happened.
Instead, familiar details were transformed.
A workplace, hospital, classroom, or other everyday setting might appear in a dream, but it would often be combined with unrelated places, shifting viewpoints, or unfamiliar surroundings. Different parts of a person's life could merge into a single vivid scene.
These findings suggest that dreaming is not a simple replay system. The sleeping mind appears to reorganize pieces of reality, combining memories with imagination, expectations, and possible future experiences.
The result can be a completely new scenario that feels immersive, emotionally intense, or surreal.
Personality May Influence Dream Vividness
The way dreams were transformed also differed from person to person.
People who were more likely to experience mind-wandering while awake tended to report dreams that changed quickly and felt more fragmented. Their dreams often moved rapidly between scenes, ideas, or perspectives.
By contrast, participants who placed greater value on dreams and believed that dreaming had personal meaning tended to describe richer and more immersive experiences. Their dream reports contained stronger perceptual detail, making the scenes feel more vivid and lifelike.
The findings do not necessarily show that believing in dreams directly causes more vivid dreaming. However, they reveal a strong relationship between how people think about dreams and how they experience or remember them.
The Pandemic Left a Mark on Dreams
The researchers also examined dream reports collected during the COVID-19 lockdown by scientists at Sapienza University of Rome. They compared those reports with data gathered in later months and years by the IMT School team.
During the lockdown, dreams contained stronger emotions and more references to restrictions, barriers, confinement, and other limitations. These themes reflected the conditions people were experiencing in daily life.
As the months passed, those patterns became less pronounced. This gradual change suggests that dreams may evolve alongside psychological adjustment, particularly as people adapt to stressful or disruptive events.
In other words, dreams may provide clues about how the mind responds to major changes over time.
"Our findings show that dreams are not just a reflection of past experiences, but a dynamic process shaped by who we are and what we live through," explains Valentina Elce, researcher at the IMT School and lead author of the paper. "By combining large-scale data with computational methods, we were able to uncover patterns in dream content that were previously difficult to detect."
A New Tool for Studying the Sleeping Mind
The study also demonstrates how artificial intelligence could change dream research.
The NLP models were able to identify meaning and structure in dream reports with an accuracy similar to that of independent human evaluators. This suggests that computational tools may allow scientists to analyze much larger collections of dreams in a consistent and repeatable way.
That could create new opportunities for studying consciousness, memory, emotional processing, and mental health.
Because dreams are highly personal and often difficult to describe, they have traditionally been challenging to investigate on a large scale. AI-based language analysis may help researchers uncover patterns that would otherwise remain hidden.
This work was supported by a grant from the BIAL Foundation (#091/2020) and by the TweakDreams ERC Starting Grant (#948891). The research was conducted at the IMT School for Advanced Studies Lucca, in collaboration with researchers from Sapienza University of Rome and the University of Camerino.%!s()