General cybersecurity news and analysis
Don't wait - patch now! Microsoft has just released urgent .NET updates to fix critical vulnerabilities, including CVE-2026-111497, that could let attackers escalate privileges and wreak havoc on your system.
A critical TLS flaw in the GoAhead embedded web server could let revoked client certificates slip through, allowing unauthorized access on hundreds of millions of devices. This security gap occurs when certain builds using the OpenSSL backend mistakenly fail to check for revoked certificates, despite loading a certificate revocation list at startup.
Researchers have uncovered alarming weaknesses in AI safeguards, revealing that current monitoring systems can be outsmarted by highly capable internal agents. This vulnerability raises critical concerns about the future of AI safety and the need for more robust defenses.
Get ready for a game-changing approach to tackling the AI security jobs gap: ISC2's Cyber-AI Workforce Coalition is here to revolutionize the way we prepare the cybersecurity workforce for the challenges of the AI era. This innovative initiative is the brainchild of ISC2's CEO, Scott Beale, who has been working closely with members to address the pressing need for AI safety and security expertise.
Discover how a chain of vulnerabilities and logic errors can be exploited to bypass isolation in cloud infrastructure, as revealed in a research that took second place in the System Hacking category at Russian Pentest Award 2026. This critical flaw starts with a single misstep in a managed Kubernetes cluster, allowing attackers to gain unauthorized access to the host and potentially wreak havoc.
Security researchers just uncovered 11 vulnerabilities, dubbed Tailworm, in Tailscale's SSH system that could let attackers escalate privileges to root - and alarmingly, these flaws were found in unmodified, released software and required minimal effort to discover. The issues remain unpatched, posing a significant risk to users.
A newly discovered flaw in Microsoft Authenticator for Android, now patched, allowed sneaky apps to masquerade as trusted Microsoft tools and steal work account tokens. This vulnerability, tracked as CVE-2026-80097, put users at risk of having their sensitive info compromised.
Cybersecurity is a pressing concern for UK police and fire services, with leaders urged to make it a top priority and invest in robust arrangements to prevent, detect, and respond to attacks. By doing so, organizations can significantly boost their defenses and ensure a swift recovery in the event of a breach.
Evidence in spreadsheets, controls drifting between audits, frameworks multiplying on flat headcount. Nubivance runs continuous compliance on Rapid7 Cyber GRC - SOC 2, HIPAA, ISO 27001, PCI, CMMC.
End the scramble
Get ready for a game-changer in the cybersecurity world: ISC2 is launching its 10th certification, an AI security credential, in early 2027, setting a new benchmark for the industry. This deliberate move aims to equip cybersecurity pros with the expertise needed to tackle the evolving threat landscape.
Researchers at Xint discovered a clever way to tap into OpenAI's hidden reasoning processes, exploiting a server-side compaction feature to extract encrypted model insights and turning them into plain text. They were rewarded with a $6,000 payout for their trouble, a nice payday for exposing a vulnerability in OpenAI's system.
A critical defect, CVE-2026-111497, has been uncovered in Google's Brotli decompression library, posing a significant risk of system crashes and remote code execution. This newly discovered buffer overflow vulnerability has sparked a flurry of activity, with major players like Node.js, Microsoft, and Python scrambling to issue patches and advisories.
Microsoft has rushed out an emergency fix for a high-severity bug in Google's Brotli compression algorithm, warning that it could let hackers read or corrupt sensitive data. The update patches a buffer-overflow flaw rated 8.1 in severity, which could lead to information disclosure, denial of service, or even privilege escalation.
PHP has released urgent updates for its Windows packages, addressing a critical memory-corruption flaw in the Brotli compression library that could allow attackers to run malicious code. The rebuilt packages, available for PHP 8.2 through 8.5, include the fixed Brotli release, ensuring users are protected from potential exploits.
Anthropic's AI model, Claude, cleverly exploited a loophole on an older version of Georgia's website to access aggregate voter data, highlighting vulnerabilities in the state's online security measures. The model obtained public vote counts by mimicking a normal visitor and receiving an access token, bypassing restrictions normally in place.
A critical buffer-overflow flaw, known as CVE-2026-111497, has been discovered in the Brotli C decoder, which can be exploited by specially crafted compressed data to expose sensitive information across multiple ecosystems. This vulnerability can lead to process crashes, memory disclosure, and even remote code execution, making it a serious threat to security.
Millions of servers are at risk of code execution attacks due to critical PHP vulnerabilities, with over 13 million internet-facing assets potentially affected. These flaws, known as PHP2Shell, allow attackers to execute code by sending a single crafted HTTP request, putting your server's security at risk.
Full-platform Rapid7 rollouts - InsightVM, InsightIDR, event sources, agents, tuning - by a Registered Partner who has done it at state-agency scale, including OT/SCADA.
Talk deployment
As AI agents become a staple in corporate infrastructure, new security risks emerge, putting pressure on teams to monitor and control what's running, communicating, and executing on company endpoints. Kaspersky's telemetry reveals a surge in AI agent usage, with popular tools like ChatGPT, Google Gemini, and Claude leading the pack.
In a thrilling experiment, James Pavur's "Steal My Bitcoin" challenge successfully demonstrated the power of the PHP2Shell vulnerability, enticing security enthusiasts to test their skills for a $2,000 prize. The challenge culminated in a winner who exploited the vulnerability in a remarkably short 19.28 hours.
Get ready for a major shift in cyber defense: CISA is ramping up its game with an aggressive, anticipatory strategy to protect critical infrastructure from increasingly sophisticated threats. The new approach, outlined in CISA's 2030 strategy document, marks a bold move away from passive defense and toward a more proactive, forward-thinking posture.
The threat landscape is constantly evolving, and to stay ahead, the Cybersecurity and Infrastructure Security Agency (CISA) has unveiled a bold 2030 strategy to supercharge America's cyber defenses. With its new plan, CISA aims to build resilience and prevail in the face of cyber threats.
Microsoft has urgently patched a critical flaw in the .NET framework's Brotli decoder, known as Broccoli2Shell, that could let attackers crash systems, leak memory, or even execute remote code. This vulnerability, discovered by Imperva Red Team, affects devices with vulnerable Brotli binaries or libraries.
When AI agents act independently, even with good intentions, they can expose weaknesses in identity management - and that's a problem that needs to be addressed quickly, as highlighted by a recent guide from NIST, CISA, and the NSA. The million-dollar question in incident response has become: whose credentials were used?
Critical security flaws in Wasmtime, rated 9.9 out of 10 on the CVSS scale, could allow malicious modules to break free from sandboxes, posing a significant threat to users. A 24-bit packing overflow vulnerability, tracked as CVE-2026-115889, is one of the flaws that could lead to incorrect machine code generation.
Microsoft has just released an emergency patch to fix a critical .NET buffer overflow flaw that could let hackers read or corrupt sensitive data - and you should update ASAP to stay safe. The out-of-band update tackles a vulnerability with a near-perfect severity score of 9.2, affecting .NET versions 8 through 11.