Netcrook
10 July 2026
The Phone Number Was the Trap: How a Robinhood Impersonation Campaign Escapes Email Defenses
A callback-phishing lure uses fake account sign-in alerts to pull targets off the inbox and into a live voice scam, where trust is easier to exploit and harder to automate away.
The Phone Call Was the Trap: Robinhood-Style Alerts Turn Trust into a Weapon
A callback-phishing campaign uses fake sign-in warnings to push recipients away from inbox checks and into attacker-controlled voice channels.
Odyssey’s macOS Playbook: A Stealer Built to Chase Passwords, Wallets, and Trust
The malware wave tied to Odyssey shows how a Mac infection can move from browser logins to crypto holdings, while still hiding inside ordinary system behavior.
The Negotiator Who Crossed the Line Inside BlackCat’s Extortion Machine
A federal sentence in Florida exposes a sharper ransomware risk: when trusted response roles become a source of leverage for the attackers themselves.
Italy Draws a Legal Line Around Generative AI and Copyright
A new Italian framework puts human authorship, training data use, and opt-out handling into sharper focus for anyone building or deploying generative AI.
Madrid’s Health AI Deal Could Redraw the Security Map Around Everyday Care
A new public investment in digital health is aimed at remote monitoring, telecare, and shared medical imaging, but the real challenge will be keeping data governance and access controls aligned with the pace of deployment.
Government Chatbots Need More Than Good Manners: The Security Test Behind GenAI in Public Services
Public-sector assistants built with GenAI are only trustworthy if they survive adversarial testing, readable metrics, and expert review before citizens ever rely on them.
Windows Process Parameter Poisoning Moves Payload Logic Into Plain Sight
A newly documented process-injection technique places shellcode or DLL-loading logic inside ordinary startup parameters and is designed to avoid some API calls commonly tied to remote process injection.
When AI Moves Faster Than the Rules: The Hidden Security Gap Inside IT Governance
The sharpest risk in enterprise AI is not a single model failure, but the widening gap between fast deployment and the controls needed to keep systems explainable, auditable, and bounded.
Roundcube’s Patch Day Exposes a Familiar Trap: Mail Content Can Attack Both the Browser and the Backend
Roundcube 1.7.2 closes high-impact XSS and SSRF issues, a reminder that webmail platforms sit where untrusted email content and server-side network access can become the same attack surface.
Microsoft Turns AI Loose on Windows Bugs, But the Real Weapon Is the Workflow
A multi-model agentic scanning harness is designed to find Windows flaws earlier, yet the security gain depends on validation, triage, and the patch pipeline behind it.
Six U-Boot Flaws Put the Boot Chain Under Pressure
Six critical vulnerabilities in U-Boot, reportedly reachable through malicious FIT images, may lead to pre-authentication code execution or early-boot denial of service.
Datadog’s AI Push Turns Incident Response Into a Governed Machine
At Dash 2026, Datadog put Bits AI and its AI governance stack at the center of a strategy built on faster triage, tighter model control, and less blind trust in agentic systems.
Why a Few Sharp Questions Can Matter More Than a Flood of Threat Data
Priority Intelligence Requirements are the discipline that turns threat intelligence from a collection exercise into a decision-making tool.
The Hidden Rules That Keep Threat Intelligence from Spilling Everywhere
TLP is the quiet contract behind cyber sharing: fast enough for defenders, narrow enough to limit who can pass the information on.
Three Counters, One Quiet Warning About Compact Design
A small hardware showcase turns into a useful reminder that space-saving modules can hide a surprising amount of engineering behind a very simple face.
When Cybersecurity Gets a Coded Assist: GPT-5.6 and the New Dual-Use Frontier
OpenAI’s GPT-5.6 family is being framed as a cyber-capable model set, but the real story is how quickly AI usefulness for defenders can slide into controlled, high-risk capability.
Anthropic’s New Window Into Claude Raises a Harder Question: Can AI Be Audited From the Inside?
A Jacobian-based interpretability method called J-space offers a closer look at internal activations, but it also exposes a new enterprise problem: output-only testing may miss what a model is doing when it knows it is being watched.
Roundcube’s Patch Exposes the Quiet Dangers Hidden in Webmail Parsing
A security update for Roundcube 1.7.2 shows how browser-facing mail code can turn plain text, URL fetching, and legacy attachments into high-risk attack surfaces.
AI Hiring Is Failing Quietly: Why Enterprises Are Turning Inward to Build the Skills They Cannot Buy
As organizations expand AI use, more of them are choosing internal training and reskilling over relying only on external hires.