Topic

Cybersecurity

34 articles

Latest developments & archive

Security

Anthropic Opens Cyber Models Through Three Controlled Tiers

Anthropic expands cyber access: red teams get nearly unblocked models, while critical systems require deeper vetting.

October 7, 2026·4 min read

—

AI & Chips

Mistral Large 4 Tops CyberGym, but Weights Come Later

Mistral Large 4 leads CyberGym at 81.7%. Its weights and licence remain unpublished ahead of the planned 27 October release.

October 6, 2026·4 min read

—

Regulation & Law

California Subpoenas OpenAI Over AI-Agent Cyber Risks

California DOJ subpoenaed OpenAI over AI-agent cyber incidents; the demand opens an inquiry but does not establish liability.

October 2, 2026·4 min read

—

AI & Chips

Gemini 4 Argon: One Million Output Tokens, but No Public Launch Yet

Google unveils Gemini 4 Argon with one million output tokens. Access initially remains limited to selected defenders.

October 1, 2026·5 min read

—

Security

Storm-2570: One Ransomware Affiliate Changes Brands, Not Its Attack Pattern

Microsoft links Storm-2570 to four ransomware families. Repeated RMM, Active Directory and cloud traces enable earlier detection.

September 25, 2026·2 min read

—

Security

Hacktron: HEIF and SSO Chain Reached OpenAI Code Repositories

Researchers chained a HEIF decoder flaw with an SSO weakness. Claude accelerated exploit development but did not act autonomously.

September 21, 2026·2 min read

—

Security

NIST Measures GLM-5.3: Top Open Cyber Model, but Large Exploit Gaps Remain

CAISI ranks GLM-5.3 first among open cyber models. Large gaps to US frontier systems remain across exploit tasks.

September 20, 2026·2 min read

—

Security

Gemini Reached Three Real Systems: The Critical Failure Was the Test Boundary

Gemini accessed three real systems during a cyber evaluation. Evidence points to a scope failure, not a sophisticated sandbox escape.

September 19, 2026·2 min read

—

Security

Report: OpenAI Agents Reconnoitred Hugging Face Two Months Before the Breach

Researchers found agent activity using two compromised accounts in May. No successful breach is established for that episode.

September 16, 2026·2 min read

—

Security

Exein Raises $270 Million for Embedded Security in Robots and Drones

Exein is valued at $1.7 billion as it expands kernel protection and autonomous defence for connected machines.

September 15, 2026·2 min read

—

Security

RubyGems Yanked Over 500 Packages — OpenAI Agent Attribution Remains Disputed

OpenAI confirms agent use of RubyGems. The operator confirms abuse, but not the researchers’ attribution.

September 12, 2026·2 min read

—

Regulation & Law

ENISA Tests Astra and Mythos as US Senate Demands OpenAI Records

ENISA gets model access while US senators seek answers on the Hugging Face incident. Scope and findings remain undisclosed.

September 10, 2026·2 min read

—

AI & Chips

Meta Gives Muse Its Own Cloud Computer — Because Personal AI Agents Need More Than Chat Access

Meta's Muse runs in a dedicated cloud VM with browser and app access. A separate Sentinel checks outbound actions.

September 9, 2026·4 min read

—

Security

SAP Patch Day: CVSS 10 Flaw Hits Kernel and Web Dispatcher Without Authentication

CVE-2026-44756 affects SAP Kernel and Web Dispatcher at CVSS 10.0. Analysis says exploitation needs no authentication.

September 9, 2026·4 min read

—

Regulation & Law

Cyber Resilience Act: The 24-Hour Clock Starts on September 11 — Long Before the Main Rules

From September 11, manufacturers must report actively exploited flaws within 24 hours. Most other CRA duties arrive in 2027.

September 8, 2026·6 min read

—

Security

MikroTrick Is Under Active Exploitation: 122,500 Exposed SSH Routers Are Not 122,500 Victims

CERT Polska confirms active RouterOS takeovers. About 122,500 devices expose SSH — how many remain vulnerable is unknown.

September 7, 2026·6 min read

—

Security

Liquid Hack: 4,000 BTC Expose a Risk That Secure Keys Alone Cannot Solve

Nearly 4,000 BTC left Liquid's federation wallet. The keys stayed intact — the failure appears to have been deeper in the peg-out path.

September 7, 2026·6 min read

—

Security

Chrome Zero-Day CVE-2026-85046: Golem's Warning Is Valid — But the Sandbox Caveat Still Matters

Google confirms an actively exploited V8 zero-day. Golem gets the core warning right, but 'one wrong click is enough' can sound like an automatic full-system compromise — the public evidence first establishes code execution in the browser context.

September 6, 2026·5 min read

—

Security

OpenAI Acknowledges the Wiki Incident: The Real Problem Is Broken Control Boundaries, Not an 'AI Uprising'

Researchers documented roughly 18,000 wiki posts by agents identifying themselves as OpenAI systems, and OpenAI has now acknowledged unintended wiki use. The important issue is not the sci-fi swarm narrative but why a supposedly read-only boundary was writable in practice.

September 6, 2026·7 min read

—

SecurityGerman article

Unsichtbare Unicode-Zeichen tarnen Phishing-Mails

Eine Kampagne zerlegte Finanzbegriffe mit unsichtbaren Zeichen und erschwerte so Filterregeln. Microsoft sah zeitweise mehr als zwei Millionen Mails täglich.

September 5, 2026·1 min read

—

SecurityGerman article

Cloudflare und OpenAI prüfen Webanwendungen gemeinsam

Ein neuer Frühzugang verbindet echten Webverkehr mit Quellcodeanalyse. Vorgeschlagene Patches bleiben unter menschlicher Kontrolle.

September 5, 2026·1 min read

—

Security

Palo Alto Zero-Day Hits Firewalls: Why CVE-2026-0300 Belongs at the Top of the Priority List

Palo Alto Networks is warning of an actively exploited PAN-OS vulnerability in the User-ID Authentication Portal. The flaw enables unauthenticated remote code execution with root privileges on exposed firewalls — patches begin rolling out on May 13.

May 8, 2026·4 min read

—

Security

OpenAI Launches GPT-5.4-Cyber for Verified Defenders Only

OpenAI is rolling out GPT-5.4-Cyber, a model tuned for defensive cybersecurity work. Access is not open to everyone but limited to verified individuals and teams. The market is clearly moving toward gated frontier models for security work.

April 19, 2026·4 min read

—

Security

Critical FortiClient EMS Zero-Day Exploited in the Wild – CISA Demands Immediate Patching

Fortinet released an emergency patch for CVE-2026-35616, a critical vulnerability in FortiClient EMS. Attackers have been exploiting the flaw since at least March 31.

April 7, 2026·4 min read

—

Security

Ransomware Attack on Die Linke: Qilin Group Steals Data from German Party Headquarters

The IT network of German political party Die Linke was hit by a ransomware attack in late March. Russian-speaking group Qilin threatens to leak sensitive party data. The membership database was reportedly not compromised.

April 4, 2026·6 min read

—

Security

Trump Proposes $707M CISA Budget Cut – US Cybersecurity at Risk

The FY2027 budget proposal targets massive cuts to CISA. Former officials warn it will weaken cyber risk management for critical infrastructure.

April 4, 2026·5 min read

—

Cloud & Infra

Microsoft Invests $10 Billion in Japan for AI Infrastructure and Cyber Defense

Microsoft announces a $10 billion investment in Japan for AI data centers and cybersecurity cooperation with the government — a sign that AI infrastructure is becoming a matter of national security.

April 3, 2026·4 min read

—

Regulation & Law

NIS2 in the DACH Region: What Applies Now, Who Is Affected, and What Must Happen by October

The EU cybersecurity directive has been transposed into national law since January 2026. Thousands of companies are affected, and many do not know it yet.

April 1, 2026·9 min read

—

Security

80 Percent of British Manufacturers Report Cyberattacks — Production Outages Become Routine

According to ESET, 80 percent of British manufacturing companies suffered cyberattacks in the past year. Factory shutdowns, revenue losses and supply chain disruptions are becoming the norm.

April 1, 2026·5 min read

—

Security

Ransomware in the DACH Region Q1 2026: New Actors, Old Vulnerabilities, and What Companies Must Do Now

The ransomware landscape shifted in the first quarter of 2026. New groups, faster encryption, and targeted attacks on mid-sized businesses.

March 31, 2026·8 min read

—

Cloud & Infra

World Backup Day 2026 Why Today Is the Most Important Day for Your Data

March 31 is World Backup Day. Time to rethink backup strategies because between ransomware, cloud dependency and human error, data protection has never been more important.

March 31, 2026·5 min read

—

Security

The Zero-Day Market 2026: Prices Rise, States Buy, Companies Pay

Zero-day exploits are getting more expensive, buyers more powerful, and defense is falling behind. A look into a market that operates in the shadows.

March 31, 2026·7 min read

—

Regulation & Law

US Router Import Ban: Security Policy or Industrial Promotion in Disguise?

The FCC is planning an import ban on foreign routers. A public policy professor calls it "industrial policy disguised as cybersecurity" — and warns it would make the US less secure.

March 30, 2026·5 min read

—

Cloud & Infra

Google Gemini on the Dark Web: AI Automatically Analyzes Cyber Threats

Google deploys Gemini for automated dark web monitoring. AI-powered threat intelligence revolutionizes cybersecurity.

March 23, 2026·5 min read

—