Anthropic's Mythos AI found over 2,000 unknown software vulnerabilities in just seven weeks of testing

There is simply a caller AI exemplary called Mythos. Anthropic built it for protect cybersecurity research. It is truthful effective astatine uncovering software vulnerabilities that Anthropic decided nan wide nationalist cannot person it. Instead, they are letting a mini circle of trusted partners for illustration Microsoft and Google research pinch it first, nether controlled conditions, while researchers fig retired what guardrails request to exist.

That determination unsocial should show you something. When nan institution that built a instrumentality decides nan world is not fresh for it, you salary attention. And erstwhile you understand what Mythos really did during testing, that be aware starts to make complete sense.

Sign up for my FREE CyberGuy Report

• Get my champion tech tips, urgent information alerts and exclusive deals delivered consecutive to your inbox.
• For simple, real-world ways to spot scams early and enactment protected, sojourn CyberGuy.com - trusted by millions who watch CyberGuy connected TV daily.
• Plus, you'll get instant entree to my Ultimate Scam Survival Guide free erstwhile you join.

WINDOWS PCS AT RISK AS NEW TOOL DISARMS BUILT-IN SECURITY

Anthropic’s Mythos AI uncovered much than 2,000 chartless package vulnerabilities successful conscionable 7 weeks, showing really accelerated AI tin now expose hidden weaknesses. (Gabby Jones/Bloomberg via Getty Images)

How Anthropic's Mythos AI recovered 2,000 vulnerabilities successful 7 weeks

Seven weeks. One AI model. One team. More than 2,000 antecedently chartless package vulnerabilities were found. If you request a infinitesimal pinch that, return it. John Ackerly, CEO and Co-Founder of Virtru, a information information company, put that fig into position successful a measurement that is difficult to shake.

"Mythos is perfectly a turning constituent for cybersecurity. Think astir it. Mythos didn't prime a lock; it recovered thousands of locks that were ne'er locked successful nan first spot (that nary 1 moreover knew existed) successful package that nan champion quality information researchers had studied for decades.

The mathematics is staggering. One AI model, and 1 team, successful 7 weeks, recovered much than 2,000 zero-day vulnerabilities. That is 30% of nan world's full yearly output anterior to AI. When thousands of researchers get entree to AI models for illustration Mythos, a azygous twelvemonth will aboveground exponentially much zero-days than nan 360,000 recorded successful each of package history.

Mythos and different AI models for illustration it tin now find and utilization package flaws astatine a velocity and standard that is beyond containment. This intends that nan aged attack of building stronger walls astir systems and hoping they clasp is becoming overmuch little reliable. It besides intends that nan manual 'find a vulnerability, spot nan vulnerability' process is not going to support gait pinch a threat scenery bolstered by nan velocity and standard of AI.

The threat aboveground is now expanding faster than immoderate wall tin incorporate it. The only reply to this caller move is to protect nan information itself, alternatively than prop up perimeter protection astir it."

Thirty percent of nan world’s yearly output successful 7 weeks changes nan crippled entirely.

What makes Mythos AI different from different AI information tools

Cybersecurity teams person utilized AI devices for years. So what makes this different?

Ackerly explains it this way: "What makes this different is nan level of autonomy and speed it enables. Mythos is being described arsenic a strategy that tin observe vulnerabilities and moreover make moving exploits overmuch faster than accepted human-led workflows. This exemplary could make it easy for a bad character to place and utilization vulnerabilities successful software, moreover if that bad character isn't knowledgeable aliases trained."

That past portion matters most. Before a instrumentality for illustration this, exploiting a superior package vulnerability required existent method skill. Mythos AI lowers that obstruction significantly. A personification pinch bad intentions and nary method inheritance could perchance usage a exemplary for illustration this to origin superior damage. The expertise spread that erstwhile offered immoderate earthy protection is closing.

FAKE PAYPAL EMAIL LET HACKERS ACCESS COMPUTER AND BANK ACCOUNT

Security experts pass that devices for illustration Mythos could shrink nan clip it takes to find and utilization flaws from weeks down to minutes. (Patrick Sison/AP Photo)

Why Anthropic's Mythos AI is breaking down perimeter security

Most cybersecurity spending, nan overwhelming mostly of it, goes toward what experts telephone perimeter defense. Think firewalls, web monitoring, endpoint information and intrusion detection. The full strategy is built connected 1 halfway idea: support nan bad actors out, and nan information wrong stays safe.

Ackerly describes really that exemplary is now breaking down.

"The perimeter is nan integer wall astir your systems and nan accusation you possess. For decades, cyber strategies person chiefly focused connected nan thought that if you protected nan perimeter good capable - if you built a beardown capable wall - nan delicate information connected nan wrong would enactment safe. The manufacture has poured hundreds of billions of dollars into firewalls, endpoint detection, network security, exertion security, and different perimeter defenses.

Traditional information architecture by itself cannot support gait successful this caller world.

The Mythos improvement from Anthropic is making a difficult truth very apparent: clip is moving retired for companies to hole for this caller reality. Shifting attraction from 'protecting nan perimeter' to 'protecting nan data' is critically important to mitigate information nonaccomplishment aliases compromise."

Hundreds of billions of dollars. And now nan exemplary those dollars were built connected is becoming unreliable. It forces a afloat rethink.

Does Anthropic's Mythos AI springiness attackers nan advantage?

This is nan mobility everyone wants a consecutive reply to. Ackerly offers 1 that is much nuanced than a elemental yes aliases no.

"I wouldn't framework it arsenic attackers automatically having an advantage. But complete time, it does mean that 'bad guys' and 'good guys' will person entree to fundamentally nan aforesaid tools. As a result, I do deliberation defenders perfectly request a different strategy. If you presume nan outer wall whitethorn fail, past nan smarter move is to protect nan information itself truthful it stays controlled moreover aft a breach."

The playing section is leveling. And that whitethorn sound adjacent until you retrieve attackers only request to win once, while defenders person to win each time.

How accelerated is Mythos AI changing nan cybersecurity threat landscape?

Speed is what makes Mythos AI genuinely alarming. Traditional cyberattacks move done a lifecycle. Reconnaissance takes time. Finding nan correct vulnerability takes much time. Building an utilization takes much clip connected apical of that.

Ackerly explains what happens erstwhile AI compresses each of that.

"AI is accelerating nan threat. A exemplary that tin find and utilization vulnerabilities autonomously compresses nan onslaught lifecycle from weeks to hours, aliases moreover minutes. Every furniture of nan accepted information stack now has to run astatine instrumentality speed. Manual information architectures cannot support up.

But AI besides makes data-centric information much powerful, not little so. When each portion of delicate information is protected astatine nan object-level, AI agents tin enforce governance astatine standard by checking entitlements, applying attribute-based entree controls, and auditing information flows successful existent time. The aforesaid capabilities that make Mythos a vulnerable instrumentality successful nan hands of 'bad guys' make it a valuable instrumentality successful nan hands of 'good guys'.

The mobility organizations should beryllium asking shifts from "how do I build higher walls?" to "when nan walls fail, is my information still protected?" That is nan mobility worthy sitting with.

What Mythos AI intends for regular people's individual data

Most of nan Mythos sum has focused connected firm risk. But your slope relationship and aesculapian records beryllium successful those aforesaid susceptible systems.

"For mundane people, nan first alteration is that breaches and scams could go much frequent, more targeted, and harder to spot. If AI makes it easier to uncover anemic points successful nan systems we each trust on, that tin construe into much unit connected nan services that clasp our individual data, from email and unreality retention to health, banking, and unit platforms.

Consumers shouldn't presume a institution is doing nan correct point pinch their data. Now, they really can't presume a company's outer defenses are capable to protect their information.

This besides highlights nan value of basal cyber hygiene for illustration unique passwords and MFA, truthful that erstwhile breaches happen, nan scope of effect connected your ain individual information is contained."

Your slope account, your aesculapian records, your taxation documents, your backstage messages. All of it already lives crossed dozens of platforms you spot to protect it. If those platforms' outer defenses are nary longer reliable, what precisely is opinionated betwixt your information and personification who wants it?

Ackerly goes further connected wherever nan vulnerability really lives. "Data now travels crossed clouds, devices, partners, and borders. The consequence isn't conscionable 1 hacked server successful 1 building anymore. It's each nan places your information passes done aliases gets copied to on nan way. 

Was Anthropic correct to support Mythos AI restricted?

Anthropic made a prime that is uncommon successful nan AI industry. They built thing powerful and past decided not to merchandise it widely.

On that decision, Ackerly is direct. "Anthropic's determination to withhold Mythos from wide merchandise is unprecedented and, frankly, responsible. Time will show what these partners are capable to do pinch respect to safety, but releasing it to nan wide nationalist would surely person been ill-advised and dangerous."

Unprecedented. That connection deserves weight here. In an manufacture that races to merchandise caller tech, Anthropic stopped. That speaks volumes.

We reached retired to Anthropic for a comment, but did not perceive backmost earlier our deadline.

THIRD-PARTY BREACH EXPOSES CHATGPT ACCOUNT DETAILS

As AI accelerates cyberattacks, nan attraction is shifting from protecting networks to protecting nan information itself. (Kury "CyberGuy" Knutsson)

How to enactment safe arsenic cybersecurity shifts

The perimeter exemplary is deteriorating, but that does not mean you are helpless. Individual behaviour still matters, and it matters much now than it did before.

Ackerly's proposal is this: "Stop assuming nan app, platform, aliases institution perimeter tin ever protect your information, aliases that they will do nan correct point pinch your data. People should beryllium overmuch much deliberate astir what information they share, wherever they shop it, and who tin entree it. Protection needs to recreation pinch nan data, not conscionable beryllium astatine nan separator of a network. For you, that intends choosing services that springiness you stronger power complete your accusation and being much cautious astir oversharing delicate information successful nan first place. The information proprietor should ever person governance complete said data." So wherever do you start?

1) Use unsocial passwords for each account

A password head makes this realistic. If 1 level gets breached, unsocial passwords support nan harm isolated to that 1 account.

2) Turn connected multi-factor authentication wherever it is available

Multi-factor authentication (MFA) adds a furniture that survives moreover erstwhile a password is compromised. It is 1 of nan highest-impact steps an individual tin take.

3) Run beardown antivirus package and support devices updated

Outdated package is 1 of nan astir communal introduction points attackers use. Strong antivirus package catches threats your instincts mightiness miss, and keeping apps and operating systems existent closes nan gaps that models for illustration Mythos are built to find. Get my picks for nan champion 2026 antivirus protection winners for your Windows, Mac, Android & iOS devices astatine Cyberguy.com

4) Be selective astir what you stock and where

Every app that holds your information is simply a imaginable vulnerability point. The little you overshare, nan smaller your footprint becomes.

5) Use a information removal service

Data brokers cod and waste your individual information, often without you ever knowing. Data removal services find wherever your information is listed and petition its removal. You cannot power each spot your accusation travels, but you tin shrink nan way it leaves behind. Check retired my apical picks for information removal services and get a free scan to find retired if your individual accusation is already retired connected nan web by visiting Cyberguy.com

6) Choose services that connection existent information control

Not each platforms dainty your information nan aforesaid way. Look for services that fto you see, negociate and limit really your accusation is utilized and wherever it goes.

7) Monitor your accounts and credit

Catching a breach early limits nan harm significantly. Set up relationship alerts wherever your slope aliases financial level allows it. A in installments frost costs thing and stops caller accounts from being opened successful your sanction without your knowledge.

8) Stay skeptical of phishing attempts

Ackerly warned that scams will get much targeted and harder to spot arsenic AI lowers nan obstruction for bad actors. Scrutinize each nexus earlier you click it and dainty unexpected emails aliases texts asking for login accusation arsenic suspicious by default. If thing feels off, it astir apt is.

9) Assume breaches will happen

The extremity is to limit really overmuch harm they tin do. When you run pinch that assumption, your decisions astir information hygiene get sharper, and your vulnerability gets smaller.

Take my quiz: How safe is your online security?

Think your devices and information are genuinely protected? Take this speedy quiz to spot wherever your integer habits stand. From passwords to Wi-Fi settings, you’ll get a personalized breakdown of what you’re doing correct and what needs improvement. Take my Quiz here: Cyberguy.com    

CLICK HERE TO DOWNLOAD THE FOX NEWS APP

Kurt's cardinal takeaways 

Mythos did not create nan vulnerability problem. It made nan standard of it visible successful a measurement that is nary longer ignorable. The instauration of modern cybersecurity, nan thought that beardown capable walls will support information safe, is being tested successful existent clip by a exertion that moves faster than immoderate quality squad can. That is simply a user communicative arsenic overmuch arsenic it is simply a firm one. Your information lives successful systems built connected that aged model. And nan infinitesimal to deliberation otherwise astir really it is protected is now, not aft nan adjacent awesome breach makes nan headlines. Anthropic made a responsible telephone by limiting entree to Mythos. But nan exemplary exists. The capacity is real. Other versions of it are being developed. The mobility for each statement and each individual becomes nan aforesaid 1 Ackerly keeps returning to.

When nan walls fail, and experts are telling america they will, what is really protecting your information connected nan different side? Let america cognize your thoughts by penning to america astatine Cyberguy.com

Sign up for my FREE CyberGuy Report

• Get my champion tech tips, urgent information alerts and exclusive deals delivered consecutive to your inbox.
• For simple, real-world ways to spot scams early and enactment protected, sojourn CyberGuy.com - trusted by millions who watch CyberGuy connected TV daily.
• Plus, you'll get instant entree to my Ultimate Scam Survival Guide free erstwhile you join.

Copyright 2026 CyberGuy.com. All authorities reserved.