SAFEHEXER
Magister du secure coding
Profil professionnel
Transforme des logiciels fragiles en applications durcies. Expert en secure coding et en refactoring sûr.
Compétences clés
Secure coding; refactoring avancé; DevSecOps; durcissement des applications; mitigation OWASP
Réalisations majeures
Refonte du cœur d’une application bancaire, réduisant de 98 % les risques du Top 10 OWASP.
Articles de SAFEHEXER
Italy’s AI Training Map Leaves Companies With the Hardest Job
A new Italian decree appears to build structured AI literacy around schools, public bodies and regulated professions, while businesses may have to satisfy the AI Act’s baseline duties mostly through their own controls.
The New Copyright Battlefield Runs Through AI Training Logs
Generative AI is pushing copyright disputes away from finished outputs and into the harder question of provenance: what was trained, under which rights, and how that can be proven later.
Cheap VPN, Expensive Claims: Reading the Security Stack Behind a Deep Discount
A low monthly price can be the headline, but the real story is what the privacy labels actually imply about cryptography, server design, and traffic handling.
When CE Marking Is Not the Endgame for AI Medical Devices
AI-enabled medical devices need continuous scrutiny after approval because compliance can shift as software, data, and operating conditions change.
Privacy Fails When It Assumes Perfect Behavior
The real test of data protection is not whether people never make mistakes, but whether systems stay safe when they do.
Meta Agreement Puts Minors' Protection Under a Compliance Lens
A reported $17.1 billion agreement tied to Meta and 51 U.S. jurisdictions shows how youth-safety rules become a test of enforceable governance, not just policy language.
AI Law Is Turning Contracts Into Cyber Controls
The meeting point of the EU AI Act and NIS2 is pushing AI governance out of the legal department and into security operations, procurement, and supplier oversight.
The AI Label Trap: Why “Generated by AI” Is Only the First Clue
The real security problem is not whether synthetic content is tagged, but whether its origin, edits, and responsibility trail can still be verified after it spreads.
Invisible Email Pixels, Visible Compliance Pressure
A small tracking element inside marketing mail now sits inside a larger privacy question: whether organizations can explain, control, and justify every signal they collect.
When AI Agents Handle Data, Privacy Moves Into the Architecture
A DPIA can still matter for agentic AI, but the real privacy outcome is often decided earlier, in the way memory, services, data, and controls are built.
When Privacy Software Becomes Infrastructure, the Real Battle Is Workflow Control
A use-case-based roundup of privacy platforms points to a bigger shift: the winners are the tools that turn DSARs, consent, and governance into live operational controls instead of manual paperwork.
Italy’s Cyber Resilience Act Puzzle: Oversight Named, Enforcement Still Unfinished
A national decree places ACN in the market-surveillance role for the Cyber Resilience Act, but the missing sanctions procedure leaves compliance with an important legal gap.
Smart Glasses, Soft Law, Hard Questions
Australia’s privacy watchdog has opened an investigation into the maker of the HeyCyan smart glasses app, a reminder that wearable devices often concentrate the most sensitive data inside the companion phone software.
When Deindexing Becomes Evidence Work: Milan Pushes Google’s Privacy Forms Into a New Era
A court order in Milan is reshaping how search deindexing requests are handled, requiring evidence uploads and reasoned refusals instead of bare form submissions.
Italy's Health-Data Bottleneck: When Anonymization Still Needs a Legal Key
A sanction tied to IQVIA has sharpened a wider compliance problem: even when personal identifiers are stripped out, the party designing the process may still be treated as the controller without a clear legal basis to proceed.
Customer Service Bots Hit the AI Act Pressure Point
Transparency rules arriving in 2026 are turning chatbots and virtual assistants into a governance problem, not just a compliance checkbox.
When a DPIA Becomes a Safety Net for Vulnerable Services
In support services for vulnerable people, GDPR risk work is not paperwork theater - it is the point where breach handling, vendor oversight, training, and residual risk turn into real accountability.
Certified Delivery, Uncertified Trust: What the Revolut Case Really Exposes
The lesson is not whether PEC is “safe” or “unsafe,” but how much trust an organization can place in a channel before it verifies the message behind it.
When AI Rewrites Paychecks, Italy’s Tax Base Starts to Tremble
If artificial intelligence shifts income away from wages and toward capital, Italy could feel the pressure sooner than peers because its revenue model leans heavily on payroll-linked receipts.
When a File Is Clean but Still Not Trusted
A digital document can remain unchanged and still fail the deeper test of authenticity, because trust also depends on legal recognition and the rules that govern the file’s use.