Red Hot Cyber

Cybersecurity, Cybercrime News and Vulnerability Analysis

Latest news

Anthropic Launches Claude Fable 5: Is the Mythos Class Model Jailbreak-Proof? innovation

Anthropic Launches Claude Fable 5: Is the Mythos Class Model Jailbreak-Proof?

Claude Fable 5 and Claude Mythos 5 are the new artificial intelligence models launched today by Anthropic, with unprecedented capabilities...
Luigi Zullo - June 10, 2026
C0XMO: Malware Infects Servers and Eliminates Other Malware Cybercrime

C0XMO: Malware Infects Servers and Eliminates Other Malware

Researchers at Fortinet have warned of a new variant of the Gafgyt botnet, dubbed C0XMO. The malware targets routers with...
Carolina Vivianti - June 9, 2026
Linux Under Fire: Just 4 Bytes Are Enough for Root Access Vulnerability

Linux Under Fire: Just 4 Bytes Are Enough for Root Access

A newly discovered vulnerability, tracked as CVE-2026-31431 and nicknamed Copy Fail, stands out as one of the most insidious Linux...
Bajram Zeqiri - April 30, 2026
Forget Alan Parsons! Apple Music and Deezer have become dumping grounds for AI-generated music innovation

Forget Alan Parsons! Apple Music and Deezer have become dumping grounds for AI-generated music

For those who love the music of Pink Floyd, they surely know Alan Parsons, the legendary sound technician who mixed...
Massimiliano Brolli - April 28, 2026
Diary of a Threat Actor: from Ghost of the Web to Unmasked Man Culture

Diary of a Threat Actor: from Ghost of the Web to Unmasked Man

Twenty-five years old, a piece of malware to remotely control dozens of machines, and a botnet as a final act...
Olivia Terragni - April 27, 2026
FABRICKED: AMD Cloud Security Flaw Exposed, Is Your Data Safe? Hacking

FABRICKED: AMD Cloud Security Flaw Exposed, Is Your Data Safe?

Cloud servers that promise complete data protection are not as impenetrable as they seem. Researchers have found a way to...
Bajram Zeqiri - April 18, 2026
Frontier AI for Defenders: CrowdStrike and OpenAI TAC innovation

Frontier AI for Defenders: CrowdStrike and OpenAI TAC

CrowdStrike has been selected for OpenAI'sTrusted Access for Cyber (TAC) program. Today, OpenAI releasedGPT-5.4-Cyber, a frontier model designed for defensive...
Carolina Vivianti - April 17, 2026
AI and Security: Is the Mythos Model a Breakthrough or Just Hype? Marcus Hutchins’s Vision Culture

AI and Security: Is the Mythos Model a Breakthrough or Just Hype? Marcus Hutchins’s Vision

With the boom in artificial intelligence, bold claims about breakthroughs in cybersecurity are becoming increasingly common, but not all experts...
Carolina Vivianti - April 16, 2026
Harvest Now, Decrypt Later: risks to your data in the near future innovation

Harvest Now, Decrypt Later: risks to your data in the near future

Article written by Massimo Dionisi translated by Tara Lee The Harvest Now, Decrypt Later (HNDL) model describes an attack strategy...
Tara Lie - April 1, 2026
From CVSS 9.8 to 7.0: What Happened to the Telegram Bug? Vulnerability

From CVSS 9.8 to 7.0: What Happened to the Telegram Bug?

Yesterday, a critical vulnerability affecting the Telegram messenger was reported in the registry of the Zero Day Initiative (ZDI), identified...
Bajram Zeqiri - March 30, 2026
Just one Telegram sticker can hack you: the critical 9.8 RCE vulnerability still has no patch Vulnerability

Just one Telegram sticker can hack you: the critical 9.8 RCE vulnerability still has no patch

There is something deeply unsettling about this vulnerability: no click is required, nothing needs to be opened. Simply receiving the...
Carolina Vivianti - March 28, 2026
The Click War: Pornhub and the Global Standoff Over Age Verification Cyberpolitics

The Click War: Pornhub and the Global Standoff Over Age Verification

The digital landscape is undergoing a radical transformation, pitting adult industry titans against national regulators. At the heart of this...
Carolina Vivianti - March 25, 2026
   Next »
Claude Fable 5 e Claude Mythos 5, due nuovi modelli di intelligenza artificiale lanciati da Anthropic, offrono capacità senza precedenti nel settore della cybersecurity. Fable 5, appartenente alla classe Mythos, è reso sicuro per l'uso generale e supera tutti gli altri modelli disponibili al pubblico in termini di performance su benchmark di capacità AI. Eccelle in software engineering, knowledge work, visione, ricerca scientifica e altre aree, con un vantaggio crescente man mano che il compito diventa più lungo e complesso. Questi modelli rappresentano un passo avanti significativo per l'industria della cybersecurity e l'elaborazione di compiti avanzati.

Anthropic Launches Claude Fable 5: Is the Mythos Class Model Jailbreak-Proof?

Claude Fable 5 and Claude Mythos 5 are the new artificial intelligence models launched today by Anthropic, with unprecedented capabilities in the cybersecurity sector. Fable 5 is a model of...
Un robottino bianco in stile cyberpunk, con cingoli neri e occhi rossi e arrabbiati, si trova in una sala server buia. Dal suo braccio-cannone spara un raggio laser rosso che distrugge un rack di server, facendolo esplodere in mille pezzi con fumo e scintille. Un display sul server colpito mostra un teschio e la scritta "SYSTEM COMPROMISED", mentre un elenco a destra segnala che i server da 07 a 10 sono offline. In alto a sinistra, un'interfaccia olografica rossa intitolata "BOT.EXE" mostra lo stato "FURIOUS" e la modalità "DESTRUCTION".

C0XMO: Malware Infects Servers and Eliminates Other Malware

Researchers at Fortinet have warned of a new variant of the Gafgyt botnet, dubbed C0XMO. The malware targets routers with DD-WRT firmware, but researchers have also identified versions for ARM,...

Linux Under Fire: Just 4 Bytes Are Enough for Root Access

A newly discovered vulnerability, tracked as CVE-2026-31431 and nicknamed Copy Fail, stands out as one of the most insidious Linux kernel bugs identified in recent years. The issue originates in...

Forget Alan Parsons! Apple Music and Deezer have become dumping grounds for AI-generated music

For those who love the music of Pink Floyd, they surely know Alan Parsons, the legendary sound technician who mixed masterpieces like The Dark Side of the Moon, as well...
GhostExodus Diary of a Threat Actor

Diary of a Threat Actor: from Ghost of the Web to Unmasked Man

Twenty-five years old, a piece of malware to remotely control dozens of machines, and a botnet as a final act of desperation to take justice into his own hands. Then,...
Un'immagine concettuale e drammatica in primo piano mostra un processore AMD posizionato al centro, che sembra esplodere o frantumarsi sotto una pressione immensa. Il chip, di forma quadrata con i bordi metallici tipici dei processori moderni, presenta il celebre logo "AMD" in bianco su uno sfondo nero opaco. La superficie del processore è attraversata da profonde crepe realistiche che suggeriscono una rottura strutturale imminente. Attorno al componente elettronico, centinaia di frammenti di ghiaccio cristallino o vetro si irradiano verso l'esterno, creando un effetto di esplosione radiale. Le schegge sono traslucide, con bordi taglienti che riflettono una luce fredda e soffusa, contrastando nettamente con la superficie scura e ruvida della base sottostante, che ricorda il cemento o la pietra scura. L'illuminazione è d'atmosfera, con toni freddi che spaziano dal blu notte al grigio antracite, conferendo all'immagine un tono tecnologico ma catastrofico. I dettagli sono estremamente nitidi, dalle venature dei frammenti di ghiaccio ai piccoli contatti dorati visibili lungo il bordo inferiore del chip. L'insieme comunica visivamente concetti come il superamento dei limiti di potenza, il surriscaldamento estremo (paradossalmente rappresentato dal ghiaccio che si spezza) o il fallimento critico di un sistema hardware ad alte prestazioni.

FABRICKED: AMD Cloud Security Flaw Exposed, Is Your Data Safe?

Cloud servers that promise complete data protection are not as impenetrable as they seem. Researchers have found a way to bypass one of the main security technologies of AMD processors,...

Frontier AI for Defenders: CrowdStrike and OpenAI TAC

CrowdStrike has been selected for OpenAI'sTrusted Access for Cyber (TAC) program. Today, OpenAI releasedGPT-5.4-Cyber, a frontier model designed for defensive cybersecurity, and expanded the TAC program to give verified, selected...
Primo piano di Marcus Hutchins davanti a una postazione da lavoro, con sfondo composto da pannelli esagonali decorativi e monitor spenti. Indossa una felpa chiara e un microfono clip. Sopra l’immagine è sovrapposta la scritta grande “ANTHROPIC”, suggerendo un contesto legato all’intelligenza artificiale e alla tecnologia.

AI and Security: Is the Mythos Model a Breakthrough or Just Hype? Marcus Hutchins’s Vision

With the boom in artificial intelligence, bold claims about breakthroughs in cybersecurity are becoming increasingly common, but not all experts are willing to accept them at face value. A new...
Harvest Now, Decrypt Later

Harvest Now, Decrypt Later: risks to your data in the near future

Article written by Massimo Dionisi translated by Tara Lee The Harvest Now, Decrypt Later (HNDL) model describes an attack strategy where encrypted data is collected today – despite being indecipherable...
Il logo di Telegram contornato da tanti stikers in una atmosfera festosa e armoniosa.

From CVSS 9.8 to 7.0: What Happened to the Telegram Bug?

Yesterday, a critical vulnerability affecting the Telegram messenger was reported in the registry of the Zero Day Initiative (ZDI), identified as ZDI-CAN-30207. The issue was discovered by Michael DePlante, a...

Discover the latest critical CVEs issued and stay updated on the most recent vulnerabilities. Or search for a specific CVE