From BleepingComputer · 40 stories
OpenAI's Astra AI Model Solves 10 Long-Standing Math and Computer Science Problems
OpenAI introduced Astra, an unreleased AI model, after an internal version achieved ten significant advances in mathematics and theoretical computer science. These problems had seen no progress for at least a decade, and in some cases, much longer, indicating a potential shift in AI's capability for foundational scientific research.
RefluXFS Linux Kernel Flaw Allows Local Root Access on XFS Filesystems
A nine-year-old Linux kernel flaw, dubbed RefluXFS (CVE-2026-64600), allows unprivileged local users to overwrite root-owned files on XFS filesystems and gain persistent root access. This race condition vulnerability affects systems running Linux kernel v4.11 or later with XFS filesystems created with `reflink=1`, including default installations of Red Hat Enterprise Linux, Fedora Server, and Amazon Linux.
Fireworks AI discloses new breach after cloud provider hack, impacting customer data
Fireworks AI, an AI infrastructure startup, disclosed a new security breach where attackers accessed its development machines by compromising a third-party cloud provider. This incident, occurring on October 23, led to the exposure of customer resource metadata, billing information, and incomplete model weights and datasets. This follows an earlier breach in early October where API keys and tokens were accessed.
Chainalysis discloses data breach from "offensive agentic techniques"
Blockchain intelligence firm Chainalysis experienced a security incident where attackers gained unauthorized access to its internal systems and tools using "offensive agentic techniques." Customer information, including names, email addresses, user IDs, and organizational roles, was accessed, but no evidence suggests sensitive investigation data or authentication credentials were exposed. The company is working with cybersecurity experts and using LLMs for a security review.
Reddit to Require User Login for Site Access, Starting with International Tests
Reddit will transition to a "logged-in-only experience," requiring users to log in to browse the site. This change, announced by CEO Steve Huffman, aims to protect Reddit's data from unauthorized scraping, improve ad targeting, and enhance user engagement. Testing will begin in select markets outside the US in the coming months.
ConnectWise Credential Compromised, Granting Access to 173 Microsoft 365 Tenants
A compromised credential for ConnectWise Cloud Backup (formerly SkyKick) has been used by unknown threat actors to access 173 Microsoft 365 customer tenants. This incident allows attackers to obtain Microsoft Graph tokens and access sensitive data due to the application's broad permissions.
Claude Opus 5.5 Released, Offering Improved Performance and Reduced Cost
Claude Opus 5.5, the first model in the Claude 5.5 family, has been released, performing at the level of Claude Fable 5.1 on most tasks while costing 40% less than Opus 5. This new model demonstrates significant performance improvements in complex tasks like code migration and software optimization, alongside enhanced safety features.
Zero-Click Candiru Spyware Delivered via Targeted Online Ads to Al Jazeera Journalists
Citizen Lab researchers confirmed that Candiru spyware infected devices at Al Jazeera through zero-click targeted online advertisements. This marks the first public technical confirmation of mercenary spyware using targeted ads for infection, requiring no user interaction. The attacks affected at least 45 Al Jazeera journalists and also targeted networks of government agencies and international organizations.
Google suspends open-source bug bounty program due to invalid AI-generated submissions
Google has suspended its Open Source Software Vulnerability Reward Program (OSS VRP) for product vulnerabilities, effective October 1, 2024. This action was taken due to an overwhelming number of invalid bug reports generated by AI, which burdened engineers and maintainers. The suspension highlights a growing challenge in cybersecurity as AI-driven tools generate false positives, impacting resource allocation for legitimate security work.
Kiteworks Urges Customers to Shut Down Servers Due to Imminent Cyberattack Threat
Kiteworks advised its customers to shut down their systems after receiving credible threat intelligence from law enforcement about an imminent cyberattack. The company recommended a precautionary shutdown to protect against potential zero-day exploits, though no compromise has been confirmed.
GitHub Actions and Pages Experience Degraded Availability, Migration to Azure Accelerated
GitHub experienced degraded availability for GitHub Actions and Pages on August 6, leading to failing or delayed workflow runs and impacting services like Copilot and GitHub Enterprise Importer. In response, GitHub is accelerating its architectural roadmap for Actions, including a full migration of the service to Azure to improve isolation, resiliency, and scalability.
Cyberattack Exposes Data of 8.7 Million Customers at Three UK Airports
Manchester Airports Group (MAG) reported a cyberattack affecting approximately 8.7 million customers across Manchester, London Stansted, and East Midlands airports. The breach compromised personal data such as email addresses, phone numbers, vehicle registrations, and postcodes, but no financial information was accessed.
Android 17 Integrates Encrypted Client Hello for Enhanced Web Privacy
Android 17 introduces platform-wide support for Encrypted Client Hello (ECH), a new internet standard that encrypts website destination information. This prevents internet service providers (ISPs) and Wi-Fi operators from seeing specific domain names, thereby limiting user profiling and targeted advertising. This makes Android the first major mobile operating system to implement ECH at a platform level.
New Android Car Head Unit Malware Uses Built-In Updaters for Ad Fraud and Botnets
A new malware family targeting Android-based vehicle head units from DoFun has been discovered, spreading through the devices' built-in update mechanisms. This malware aims to facilitate ad fraud and create a proxy botnet, marking the first documented case of malware on car head units with a device-specific infection chain.
Polish Energy Plant Cyberattack Used Novel Private APN Vector, Shutting Down Turbine
A previously undisclosed cyberattack in December 2025 targeted a small Polish combined heat and power (CHP) plant, causing a temporary shutdown of its steam turbine and water treatment system. The attack, which threatened heat supply to 50,000 residents, utilized a private Access Point Name (APN) as an attack vector, marking the first documented real-world use of this method to access industrial control systems.
Malicious `proc-macro1` crate leads to supply chain attack on `arrayref` and other Rust crates
The Rust Security Response Team identified and removed several malicious crates, including `proc-macro1`, which was used in a supply chain attack to compromise popular crates like `arrayref`, `internment`, and `append-only-vec`. This incident highlights the vulnerability of software supply chains to malicious package injections and necessitates developers to verify their dependencies.
WhatsApp Rolls Out Optional On-Device Scam Alert Feature in Limited Beta
WhatsApp has launched a limited beta of "Scam Alert," an optional feature that uses an on-device machine learning model to identify suspicious messages from unknown contacts. This feature processes message content locally on the user's device to maintain end-to-end encryption while alerting users to potential scams. Users can block, report, or ignore flagged messages, and can opt to share message data to improve the model's accuracy.
Zimbra Releases Critical Security Patches for Classic Web Client
Zimbra has released version 10.1.19 to patch a critical stored XSS vulnerability in its Classic Web Client, which could allow attackers to execute malicious code via crafted emails. Additionally, Zimbra version 10.1.20 addresses multiple vulnerabilities, including command injection and mail forwarding bypass. The updates are crucial to maintain security for users of the Zimbra Collaboration Suite.
Coca-Cola's Fairlife Hits U.S. Production Halt Due to Anubis Ransomware Attack
A ransomware attack by the Anubis group has forced Coca-Cola's Fairlife to suspend U.S. production. Hackers claim they extracted 1 TB of data, threatening to release it unless a ransom is paid. The incident raises concerns about cybersecurity in the food and beverage sector.
Microsoft Issues Record 570 Security Patches, Including Three Zero-Days
Microsoft's July 2026 Patch Tuesday included a record 570 security patches, with three zero-day vulnerabilities addressed. The increase is partly due to AI-assisted discovery, highlighting a trend in vulnerability identification and remediation.
Critical KVM/x86 Vulnerability Allows VM Escape to Host on Intel and AMD
Januscape, a 16-year-old use-after-free vulnerability (CVE-2026-53359) in Linux's KVM hypervisor, allows guest VMs to execute arbitrary code on host systems, compromising host security in multi-tenant environments. Discovered by Hyunwoo Kim, this first-known architecture-independent exploit has been demonstrated in Google's kvmCTF. Cloud providers like Google Cloud and AWS may be particularly vulnerable, posing risks of data breaches.
19-Year-Old Extradited to U.S. for Role in Scattered Spider Hacks
Peter Stokes, a dual U.S. and Estonian citizen, was extradited from Finland to the U.S. to face charges related to hacking activities with Scattered Spider. Notable incidents include a 2025 cyberattack on a luxury jewelry retailer demanding an $8 million cryptocurrency ransom. The extradition underscores efforts to combat global cybercrime.
Opera Rolls Out 'Paste Protect' to Counter ClickFix Cyber Attacks
Opera has introduced 'Paste Protect', a new feature to block malicious clipboard commands copied from websites, aimed at mitigating ClickFix-style attacks. These attacks leverage social engineering tactics to mislead users into pasting harmful code into terminals. The new feature marks the first native defense against this rising cyber threat in a major web browser.
JPMorgan, Visa, Mastercard Disclose Third-Party Data Incidents After N0n Extortion Claims
JPMorgan Chase, Visa, and Mastercard have reported data incidents involving third-party service providers, following claims by the N0n extortion group that it stole customer data. The companies state that their core systems and payment data were not compromised, but some customer names, email addresses, and marketing data may have been accessed.
Canadian Extradited to US for Alleged Cryptomining Hacks on University Supercomputers
Ryan James Roach, a 35-year-old Canadian, has been extradited to the US to face federal charges for allegedly orchestrating a cyber campaign that hijacked university supercomputers to mine cryptocurrency. From June 2016 to December 2019, Roach and co-defendant Mathiew James Stubbings, 41, allegedly compromised systems at academic institutions in New York, Massachusetts, North Carolina, and Tennessee, stealing credentials and deploying mining software. Stubbings' extradition proceedings are ongoing in Canada.
ConnectWise Disables ScreenConnect Extensions Amid Active Attacks on MSPs
ConnectWise is investigating an active malware campaign targeting Managed Service Providers (MSPs) that uses compromised ScreenConnect environments to deploy rogue remote-access clients. Cybersecurity firm Huntress identified six compromised MSPs and 286 affected endpoints across 47 organizations. ConnectWise has disabled five third-party ScreenConnect extensions across approximately 400 partner environments as a precaution, stating there is no evidence of a compromise to its infrastructure or a vulnerability in ScreenConnect itself.
New SignDrop Zero-Day Exploit Grants SYSTEM Privileges in Windows CLFS Driver
Security researcher Abdelhamid Naceri (MSNightmare) disclosed a new Windows zero-day exploit, SignDrop, which allows attackers to gain SYSTEM privileges. The exploit abuses a log-authentication feature introduced by Microsoft in October 2025 for the Common Log File System (CLFS) driver, affecting Windows 11 25H2, 26H1, and Windows Server 2025. This vulnerability can bypass antimalware self-defense mechanisms by forcing privileged processes like Windows Defender to create authentication files with SYSTEM permissions.
Notion AI Chat Transcripts Exposed in Fireworks AI Security Incident
Notion announced that less than 1% of its workspaces were affected by a security incident at its vendor, Fireworks AI. The breach exposed unattributable Notion AI chat transcripts, which did not contain user or workspace identification, nor tool call results. Notion has disabled the affected 'Insights' feature and ceased using Fireworks AI for open-weight models.
Railway Detects Unauthorized Access to Encrypted Customer Environment Variables
Railway detected unauthorized access to systems storing customer environment variables, potentially exposing encrypted and some decrypted contents like API keys and database credentials. The company believes this is part of a broader attack targeting platform companies and AI inference providers, prompting a recommendation for all users to roll their environment variables.
Qubole Customer AWS Accounts Breached via Compromised Vendor Credentials
An attacker exploited compromised credentials for a Qubole AWS account to breach multiple customer AWS environments between October 14 and October 26. The attacker assumed IAM roles within customer accounts, including unused ones, to steal stored secrets and data. This supply chain attack highlights the risks associated with third-party vendor access and persistent, often forgotten, cloud permissions.
DreamHost Initiates Mandatory Password Reset for Affected Email and FTP/SFTP Users
DreamHost is requiring a password reset for some email and FTP/SFTP users due to an active incident. Account owners can reset passwords via the control panel, and emails with instructions are being sent in batches.
Polymarket Halts Trading After Third-Party Breach, $2.5M in Crypto Traced
Prediction market platform Polymarket temporarily suspended trading after a third-party service compromise led to the distribution of crypto-draining JavaScript. Approximately $2.5 million in assets were traced to a suspected attacker's wallet, and Polymarket has stated it will reimburse all affected users.
Node.js Updates Fix Brotli Decoder RCE Vulnerability Across Supported Versions
Node.js released security updates for versions 22.x, 24.x, and 26.x to address 10 flaws, including a high-severity remote code execution (RCE) vulnerability in its bundled Brotli decompression code. This flaw, an out-of-bounds write, could allow an attacker to run arbitrary code if an application decompresses attacker-controlled Brotli data. Users of unsupported Node.js versions (20.x and earlier) must upgrade to a supported line to receive the fix.
Unpatched Tailscale SSH Flaws Allow Root Access on Other Servers From a Hacked Device
A security researcher identified unpatched vulnerabilities in Tailscale SSH that allow a compromised device to gain root access on other servers within the network. The attack requires Tailscale SSH to be configured for port forwarding and an attacker to control an initial device with network access to the target, posing a risk to networks using the affected configurations.
Eye Security Disables Vulnerable Time4Popcorn Updater to Prevent Botnet Formation
Eye Security researchers took control of abandoned domains used by the "Time4Popcorn" app, a fork of Popcorn Time, to push an update that disabled its updater service. This action prevented potential remote code execution (RCE) attacks on millions of Windows, macOS, and Android devices still running the outdated software. The vulnerability, tracked as CVE-2026-30612, stemmed from unencrypted HTTP updates and a lack of digital signature verification, which could have allowed attackers to execute malicious code with full privileges.
LiteLLM 1.107.0 on PyPI Compromised with Credential Stealer and Backdoor
The LiteLLM Python package version 1.107.0, distributed via PyPI and Docker images, was compromised in a supply chain attack. This malicious release installed a credential stealer and a remote code execution backdoor, which collected sensitive data like SSH keys, cloud credentials, and AI tool tokens, and established a persistent daemon for further code execution. PyPI has since removed the affected release, and LiteLLM is investigating the incident, noting that stolen GitHub CLI OAuth tokens from three engineers' accounts were used to publish the malicious version.
Quest Diagnostics Reports Cyber Breach of Patient Records on Quanum Lab Services Manager
Quest Diagnostics reported unauthorized access to its Quanum Lab Services Manager platform, potentially compromising certain patient records. The company discovered the breach on Wednesday and is investigating the scope of affected data and patients, while working with federal law enforcement and cybersecurity experts.
Chrome's Safe Browsing Enhanced Protection Mode Improves Phishing and Scam Detection
Google Chrome updated its Safe Browsing Enhanced Protection mode with new features to detect suspicious and dangerous sites, and to counter ClickFix social engineering attacks. These updates aim to make users ten times safer from various online threats compared to Standard Protection by leveraging Gemini models and clipboard analysis.
Revolut Experiences Widespread Service Disruption Amidst Google Cloud Outage
Revolut, a digital bank, experienced a significant service disruption affecting multiple services globally, including card payments, transfers, and ATM withdrawals. The outage, which began on Tuesday morning, left thousands of users unable to access funds or make payments. This incident occurred concurrently with elevated errors and intermittent network failures reported by Google Cloud in its europe-west2 region, impacting various Google Cloud services and other companies.
Qilin Ransomware Suspect Arrested in Japan, Extradited to Germany
An alleged member of the Qilin ransomware group was arrested in Japan and extradited to Germany, where the individual is wanted for a ransomware attack against a logistics company. This arrest targets a core member of a prolific ransomware-as-a-service operation responsible for attacks on hundreds of organizations globally.