NIS2 requirements in Romania can be implemented in three ways: in-house, with a consultant running the process, or through CISO as a Service, where an external specialist covers the NIS2 officer role. The obligations under GEO no. 155/2024, approved with amendments by Law no. 124/2025, are identical in all three. What differs is who does the work, how long it takes and how the cost is shaped: internal time, a one-off project, or a monthly retainer.

What follows compares the three routes on the same criteria, including the uncomfortable ones. None is inherently better. The choice depends on three things you already know: how many people you have, how quickly you need the result, and who carries the programme after implementation ends.

What has to be done, whichever route you take

Before choosing who does the work, it helps to fix the list of what the work is. It is the same for everyone:

  • Scoping. Establishing whether the entity falls under the ordinance and in what capacity, essential or important, including the self-assessment of disruptive effect under art. 9.
  • Notification for registration with DNSC, within 30 days of the moment the provisions become applicable, under art. 18 para. (2). Changes to registration data are reported within two weeks or three months, depending on the category.
  • Risk level assessment using the ENIRE@RO tool, whose result, validated by DNSC, sets the applicable level: Basic, Important or Essential.
  • Maturity self-assessment of risk management measures, within 60 days of submitting the risk assessment, and annually thereafter under art. 12 para. (4).
  • The substantive measures in art. 11: governance, risk analysis, incident handling, business continuity, supply chain security, access control, cryptography, testing.
  • Management training. Art. 14 para. (2) requires members of the management body to complete accredited professional training, and para. (1) makes them liable for breaches.
  • Reporting significant incidents, with an early warning within 24 hours, notification within 72 hours and a final report within one month, under art. 15 para. (7).
  • Readiness for inspection. The supervision, verification and control rules were approved by DNSC director’s Order no. 3 of 27 November 2025, together with the risk-based prioritisation methodology.

One thing does not change in any route: liability stays with the management body. Neither the consultant nor the outsourced officer takes it on. What is outsourced is the work and the expertise, not the legal responsibility.

Route 1. In-house implementation

The organisation runs the programme with its own resources: an IT or security team, an internally designated officer, and support from legal, procurement and HR.

When it works

It works where an information security management system already exists, whether certified to ISO/IEC 27001 or built internally, and where someone’s job description includes tracking legal requirements. It also works in organisations that went through GDPR with a serious process, because part of the inventories, registers and documentary discipline can be reused.

What it actually costs

There is no fee, but there is time, and time is the resource most often missing. For a medium-sized important entity, realistic effort runs to several hundred hours spread over six to twelve months, plus management time for approvals and training. That allocation shows up as other projects postponed.

Where it stalls

Stalls rarely happen on the technical side. They happen on documentation, on interpreting an ambiguous requirement, and at the first self-assessment, where the question is not what you implemented but what you can prove. The second classic obstacle is objectivity: the team that built the systems tests exactly the assumptions it used when building them, and self-assessment scores come out systematically more generous than what can be verified.

Route 2. Implementation with a consultant

An external consultant runs the project: scoping analysis, assessment of the current state against requirements, remediation plan, documentation and knowledge transfer to the internal team. The organisation still owns the programme, but does not start from zero and does not learn on its own.

When it works

This is the right route when the deadline is short, when the organisation has capable people without experience of this particular framework, or when a security function exists that needs method rather than execution. It also fits situations calling for an independent view, requested by shareholders, by a major client or by a sector authority.

What it costs

At ProDefence, consulting and implementation start from EUR 10,000 per project, and a compliance audit from EUR 5,000 per engagement. Price varies with the number of sites, infrastructure complexity, the presence of operational technology environments and the level produced by the risk assessment.

Where it stalls

The specific risk here is the project that finishes beautifully and stops there. The consultant leaves, the documentation stays, and at the first annual self-assessment nobody remembers where the evidence register was. A well-run consulting project includes knowledge transfer and a named internal owner, not only deliverables.

Route 3. CISO as a Service, acting as the outsourced NIS2 officer

An external specialist covers the cybersecurity officer role continuously, in the model known as CISO as a Service. It is not a project with a start and an end but a standing function on partial allocation: keeping the obligations calendar, updating documentation, preparing management meetings, handling the relationship with DNSC and coordinating incident response.

When it works

It suits organisations that fall under NIS2 without having the critical mass for a full-time post, which describes most important entities in Romania. An experienced specialist on permanent staff costs considerably more than a partially allocated role, and in many organisations the actual workload does not justify a full-time hire.

It also suits organisations that have finished implementation and discovered that the hard part is only starting: the obligation repeats annually, legislation changes, suppliers change, and incidents give no notice.

What it costs

At ProDefence, the outsourced NIS2 officer role starts from EUR 800 per month, on a 12-month contract with monthly invoicing. The fee depends on organisation size, the number of critical systems and suppliers, and the applicable assurance level.

Where it stalls

An external role works only if it has access and authority. If the outsourced officer learns about a new supplier after the contract is signed, or about an incident the next day, the role becomes decorative. The mandate, the access rights and the direct channel to management are settled at the start, not at the first incident.

Compared on the same criteria

CriterionIn-houseWith a consultantOutsourced officer
Who runs the programmeThe internal teamThe consultant, with the teamThe external specialist, as a standing role
Internal time requiredHighMediumLow
Speed to first evidenceSlowFastModerate but steady
Cost shapeInternal time, no feeProject, paid onceMonthly retainer
Entry cost at ProDefenceDocumentation package, EUR 1,500 or 2,000From EUR 10,000 per projectFrom EUR 800 per month
Legal liabilityManagement bodyManagement bodyManagement body
What happens after go-liveStays with the teamStays with the team, if handover happenedContinues without interruption
Objectivity of self-assessmentWeak pointGoodGood
Fits whenYou have process and peopleYou have a deadline and peopleYou have the obligation but not the person

How to choose without asking a vendor

Three questions separate the routes better than any sales deck.

Whose job description currently includes tracking legal security requirements? If the answer is a name, in-house is realistic. If the answer is a department rather than a person, the obligation will fall between chairs.

How long would it take to prove that a declared measure actually works? If the answer is hours, you have a process. If it is days of reconstruction, you need method, which means a consultant.

Who will do next year’s self-assessment? This question dispels illusions fastest. The obligation in art. 12 para. (4) repeats annually, and a closed project does not answer it.

What all three routes have in common

Whoever does the work, two things help in every scenario.

The CysNIS platform, for records and tracking

CysNIS connects in one place the elements that in practice sit apart: the scoping assessment, the maturity level, risks, measures, plans, owners, deadlines and evidence. In-house it replaces a tool you do not have. With a consultant it keeps the programme visible to management during the project. With an outsourced officer it is the shared space where they and your team work.

The first step in any route is the indicative scoping check, by sector, service, size and the special criteria that may apply.

The NIS2 documentation package, for the written part

Most of the compliance effort is not technical but documentary: policies, procedures, registers, plans, self-assessment forms and the evidence file presented at an inspection. The NIS2 documentation package covers exactly that, with 219 documents for important entities and 229 for essential ones, the difference being the modules for the Essential assurance level, for ICT and cloud environments and for OT, ICS and SCADA environments.

It fits all three routes, for different reasons. In-house it saves months of drafting from scratch. With a consultant it shortens the project, because the conversation starts from existing documents to be adapted rather than a blank page. With an outsourced officer it becomes the base they keep current.

Prices are EUR 1,500 for important entities and EUR 2,000 for essential entities, paid once, licensed for a single organisation. ProDefence S.R.L. is not registered for VAT, so the price shown is final. The package is sold only to organisations in scope of NIS2, for their own compliance programme.

Frequently asked questions

Can NIS2 be implemented without a consultant?

Yes. The ordinance does not require a consultant. It requires the outcome: measures proportionate to risk, documented, with evidence that can be presented at an inspection. An organisation with a mature security process and a designated internal officer can implement on its own.

Does an outsourced NIS2 officer take on the legal liability?

No. Under art. 14 of GEO no. 155/2024, management bodies approve the measures, oversee their implementation and are liable for breaches. Outsourcing covers execution and expertise, not liability.

How long does NIS2 implementation take?

It depends on the level produced by the risk assessment and on the starting point. For an important entity with existing security practices, a consultant-led programme reaches a first complete set of evidence in two to four months. In-house, the usual range is six to twelve months, because the work runs alongside day-to-day operations.

What does CISO as a Service mean?

It is the model where the information security officer role is covered by an external specialist on partial allocation and a long-term contract, instead of a full-time hire. In the NIS2 context, it covers the duties of the designated officer for the relationship with DNSC and for the compliance programme.

Does the documentation package replace consulting?

No. The documents cover the form, not the content specific to your organisation. Risk analysis, scoping decisions and adapting measures to the real infrastructure still have to be done, internally or with external support. The package removes the drafting work, not the analysis.

Can the route change along the way?

Yes, and the most frequent combination is a consultant-led implementation project followed by an outsourced role that keeps the programme running. The reverse, from outsourced role to in-house, works when the organisation hires a specialist in the meantime.

Where to start

The order that saves the most time is the same in all three routes: scoping first, then risk level, then the decision about who executes. Doing it the other way round, choosing a supplier before knowing the applicable level, produces quotes that cannot be compared.

For an assessment of the current state against the requirements, ProDefence offers a NIS2 compliance analysis with a remediation plan, built on evidence by DNSC-attested NIS auditors on the ProDefence team. For the standing role, details are on the outsourced NIS2 officer page. The full list of services is in the cybersecurity services register.

Romania’s National Cyber Security Directorate (DNSC) has published the cyber maturity score of the Romanian chemical sector for CyFun® control PR.PS-05.1: 4.40 out of 5. It is one of the strongest sector results in the series of measurements carried out under the PNRR 184 project, and the control assessed is about as concrete as they come: web and e-mail filters must be installed and used.

The figure deserves careful reading, in both directions. It confirms that the chemical industry has invested in the layer of defence that stops the most common attack vectors. It says very little about the distance still to cover towards compliance with GEO no. 155/2024, and that difference costs money when an inspection arrives.

What DNSC actually measured

Under the PNRR 184 project “Creating new cybersecurity competencies for society and the economy”, 1,599 beneficiaries from the public and private sectors assessed their own cyber maturity on the SecureRO platform, using the CyberFundamentals (CyFun®) framework developed by the Centre for Cybersecurity Belgium.

Two qualifications change how the result should be read. First, this is a self-assessment, not an audit. No evidence was requested for the answers given, and field experience consistently shows that declared scores are more generous than what can be verified. Second, CyFun® is not Romania’s compliance framework. It is a useful working instrument, conceptually aligned with NIS2, but the legal obligation is measured against GEO no. 155/2024, approved with amendments by Law no. 124/2025, and against the orders issued by DNSC under it.

A 4.40 out of 5 on a single technical control is a good signal. It is not a statement of compliance.

What control PR.PS-05.1 requires

The control text is short: “Web and e-mail filters must be installed and used.” Its stated objective is to reduce the risk of malware infection, phishing attacks and data breaches through the implementation and maintenance of effective web and e-mail filtering solutions.

The word that separates a score of 3 from a score of 5 is “maintenance”. An e-mail gateway bought in 2021 and left on its default configuration ticks the box for having a solution, not for its effectiveness. Real maturity looks different:

  • filtering applied to every flow, including shared mailboxes, office@ addresses and accounts used by applications;
  • sender authentication fully configured: SPF, DKIM and DMARC in reject mode, not merely monitoring;
  • attachments and links inspected at the time of access, not only at delivery;
  • web filtering that also covers endpoints outside the corporate network;
  • rules reviewed periodically, with logs retained and actually examined, not merely generated;
  • exceptions documented, each with an expiry date and an owner, rather than accumulated on request.

In a chemical company, filtering carries an additional stake. E-mail is how orders, safety data sheets, customs documents and correspondence with suppliers of regulated substances arrive. A forged message that changes a bank account, or an altered safety data sheet, produces more than a financial loss, it produces process risk.

Where a good technical score stops

The chemical sector appears in the annexes of GEO no. 155/2024, and entities above the qualifying thresholds fall into the essential or important category. The obligations that follow are not covered by e-mail filters.

In practice, an entity in the sector has an administrative path with its own steps and deadlines:

  • self-assessment of disruptive effect, governed by art. 9 of GEO no. 155/2024, for which DNSC has published a dedicated guide;
  • notification for registration, generated with the NIS2@RO tool, signed by the legal representative and sent to [email protected] or filed at DNSC headquarters;
  • risk level assessment using the ENIRE@RO tool, whose score, once validated by DNSC, sets the applicable level: Basic, Important or Essential;
  • self-assessment of the maturity of risk management measures, required by art. 18 para. (7), using the instrument matching the resulting level;
  • the substantive measures: governance, risk analysis, incident handling, business continuity, supply chain security, access control, cryptography, management training.

That last point is what catches out companies with solid technical infrastructure. The filters work, but the approved policy, the responsibility matrix, the incident register, the tested response plan and the evidence that the management body was trained are all missing. At an inspection, the documents are what is asked for.

Three checks that reveal the real score

First: send a test message with a harmless attachment carrying a double extension to a shared company address, not to the CTO’s mailbox. If it arrives, filtering covers only the main flow.

Second: check the domain’s DMARC record. If the policy is “none”, anyone can send messages in the company’s name without them being rejected. It is the cheapest control on the list and the one most often left half-finished.

Third: ask for the list of active exceptions in the filtering solution. If nobody can produce it the same day, the control is not being maintained, whatever the self-assessment says.

Supervision is no longer hypothetical

Through Order no. 3 of 27 November 2025, the DNSC director approved the rules for supervising, verifying and enforcing compliance with GEO no. 155/2024, together with the risk-based methodology for prioritising those activities. The supervisory framework exists, is public and operates on risk criteria, which means entities with a high risk level and thin documentation reach the authority’s attention first.

Where to start

The first useful step is not buying a solution, but establishing scope precisely. The CysNis platform walks through the qualifying criteria step by step and separates legal obligations from technical recommendations.

To measure the distance between the current state and the legal requirements, the next step is a NIS2 compliance analysis with a remediation plan. Members of the ProDefence team are DNSC-attested NIS auditors, and the assessment is built on evidence rather than declarations. Organisations without an internal owner for the process can cover the role through an outsourced NIS2 officer, accountable for deadlines, documentation and the relationship with the authority. The rest of the work is listed in the cybersecurity services register.

A 4.40 out of 5 on filtering is a good starting point. The rest still has to be demonstrated.

Technical filtering does not stop everything: the messages that get through reach people. The free interactive phishing exercises on CysGame (in Romanian) and the courses on CysEdu (in Romanian) complement technical controls with awareness.

Data source: DNSC, “Cyber maturity score for Romania’s chemical sector, CyFun® control PR.PS-05.1”, 11 September 2026.

The CysNIS platform, built by ProDefence for coordinated NIS2 implementation, has been ranked 9th in the IT & Tech category at INNOVENTURE 2025-2026, Romania’s National Innovation Competition for SMEs, with a final score of 3.85.

The recognition matters to us for a specific reason: the category was contested by software companies with hundreds of employees, and the entry assessed is not a general-purpose suite but a tool built for a single problem, NIS2 compliance in Romania.

The official IT & Tech ranking

RankCompanyProjectScore
1Egnosis SRLBiobank.ro4.32
2Expertware S.R.L.SIEMBIOT4.30
3Arhiv360Archive management system4.18
4Wolfpack Digital SRL3D2Cut4.15
5Planograma S.R.L.Planograma4.10
6LinnifyAgentLens4.06
7Maya InterconsCOMPLEXIO4.00
8AMC Websoft SRLCRM AMC3.91
9ProDefenceCysNIS platform3.85
10AROBS Transilvania SoftwareNC4F3.62

Assessment ran in two stages. In the first round, a team of 20 experts, five per industry, selected ten innovative companies in each sector. A committee of three leaders of Romanian innovation then decided the grand prizes across all industries. The competition is an initiative of Loop Operations, and the full ranking is published on the organiser’s site.

What CysNIS actually is

CysNIS is the platform an organisation uses to check, on an indicative basis, whether it falls under NIS2, to manage the implementation of its measures and to document compliance in one place, instead of a folder of files scattered across several departments.

It connects the elements that in practice sit apart: the scoping assessment, the maturity level, identified risks, security measures, implementation plans, owners, deadlines and the evidence supporting every measure claimed.

The problem is not reading the law. It is coordination.

Organisations falling under GEO no. 155/2024, approved with amendments by Law no. 124/2025, discover quickly that the legal text can be read in an afternoon. What consumes the months that follow is something else: who owns each measure, in what order they are implemented, what evidence supports each claim, and how to keep track of a programme spanning several departments and several suppliers.

The cost of poor coordination shows up at fixed moments written into the ordinance:

  • the annual maturity self-assessment of risk management measures, which art. 12 para. (4) requires to be submitted to DNSC year after year, a deadline missed quietly, because nobody has it in a calendar;
  • notification of a significant incident, with an early warning within 24 hours, notification within 72 hours and a final report within one month under art. 15 para. (7), deadlines that leave no time for hunting through documents;
  • supervision by the authority, whose rules were approved by DNSC director’s Order no. 3 of 27 November 2025, together with the risk-based prioritisation methodology.

In all three, the question is the same: where is the evidence. An organisation that keeps its records inside a process answers in minutes. One that keeps them in e-mails and spreadsheets spends, at best, a day reconstructing them.

What the platform does not do

Worth saying plainly, because compliance attracts a lot of promises: CysNIS does not replace legal analysis, does not replace an audit and does not stand in for a specialist’s judgement. The scoping check it offers is indicative, and the platform does not substitute for DNSC’s own procedures or official channels.

What it provides is the operational frame: organising activities, letting teams work together, tracking deadlines and preparing evidence. The scoping decision, the evidence-based assessment and the documentation presented at an inspection remain human work, at ProDefence, the work of DNSC-attested NIS auditors.

Why it matters for a small security firm

In a market where “compliance platform” has become a label attached to any checklist spreadsheet, an assessment by an independent jury, on innovation criteria, provides an outside reference point. It is not a product audit and it guarantees results for nobody. It is, however, confirmation that the direction, turning a legal text into a trackable operational programme, is considered useful by people outside the company.

For clients, the practical signal is different: the platform they rely on is not a project abandoned between two contracts, but a product under continuous development, with reputational stakes for whoever builds it.

What comes next

The INNOVENTURE 2026 gala is scheduled for 24 November 2026 at the Sheraton Bucharest Hotel, in the Platinum Ballroom, where the competition’s grand prizes will be announced.

Our thanks to the organisers and the jury. Development continues, with the emphasis on usefulness and traceability. The rest is marketing detail.

Where to start

If you do not yet know whether your organisation falls under NIS2, the first step is the indicative scoping check on CysNIS, by sector, service, size and the special criteria that may apply.

If scope is already clear and the hard part is next, the starting point is a NIS2 compliance analysis with a remediation plan, built on evidence. Organisations without an internal owner to carry the programme forward can cover the role through an outsourced NIS2 officer, including the annual deadlines and the relationship with the authority. The full range of services is in the cybersecurity services register.

Alexandru Angheluș has published “Transformation”, a work examining artificial intelligence beyond its conventional role as a technological application and analysing it as an emerging cognitive layer positioned between human beings and reality.

Artificial intelligence is increasingly involved in filtering information, generating content, recommending options, interpreting complex environments and assisting human and institutional decision-making.

Against this background, Transformation starts from a fundamental question: not whether artificial intelligence will change the world, but whether that transformation will remain intelligible, governable and compatible with the material limits of Earth.

The work approaches AI from a multidisciplinary perspective that connects technology, human cognition, organisational transformation, cybersecurity, governance, geopolitics and the physical resources required to sustain increasingly complex artificial intelligence infrastructures.

Artificial intelligence as a cognitive layer

AI systems no longer operate exclusively as isolated technological tools. They increasingly mediate the relationship between individuals, organisations and the environments in which decisions are made.

Search engines prioritise information. Recommendation systems influence choices. Generative systems create text, images, software and synthetic media. Decision-support platforms assist organisations in evaluating risks, allocating resources and defining courses of action.

As this mediation expands, artificial intelligence gradually becomes part of the cognitive infrastructure through which reality is perceived, interpreted and acted upon. Transformation explores the consequences of this transition.

The central issue is therefore not simply the technological performance of artificial intelligence systems, but the changing relationship between capability, vulnerability and autonomy.

The cognitively assisted Earth

One of the concepts introduced in the work is the cognitively assisted Earth – a state in which artificial intelligence systems become increasingly integrated into the operational environment of human and institutional life.

However, artificial intelligence is not immaterial. Its development and operation depend on physical infrastructures and resources, including:

  • energy;
  • water;
  • computing infrastructure and data centres;
  • semiconductors and critical minerals;
  • telecommunications infrastructure;
  • large volumes of data;
  • specialised human labour.

The expansion of artificial intelligence must therefore also be understood as a material transformation. The question of AI development cannot be separated from energy availability, supply-chain resilience, access to strategic resources, environmental constraints and geopolitical competition.

The cognitively assisted Earth is consequently both a technological and a physical system.

HumanAI – the artificially mediated human

A second analytical concept developed in Transformation is HumanAI. The notation describes the individual whose capabilities, vulnerabilities and decision processes are increasingly mediated by artificial intelligence systems.

AI can significantly extend human capability. It can improve access to information, accelerate analysis, support learning, automate repetitive activities and assist decision-making in highly complex environments.

At the same time, increased reliance on artificial systems introduces new vulnerabilities. When individuals delegate information retrieval, interpretation or decision-making to AI systems, their autonomy can either increase or decrease. The direction of that transformation depends on several factors, particularly competence, degree of delegation and mechanisms of verification.

A person capable of understanding how AI systems operate, evaluating their outputs and independently verifying critical information may use artificial intelligence as a powerful cognitive amplifier. Without those safeguards, the same technology may produce dependency, cognitive outsourcing or an illusion of informed decision-making.

OrgAI – the artificially mediated organisation

The same transformation is taking place at organisational level. Transformation introduces the concept of OrgAI to describe organisations whose operational capabilities, vulnerabilities and decision processes are increasingly mediated by artificial systems.

AI is progressively entering areas such as risk assessment, cybersecurity operations, financial analysis, logistics, intelligence, compliance, customer interaction, strategic planning and executive decision support. This changes organisational capability, but also the nature of organisational risk.

An organisation that delegates critical analytical or operational functions to artificial intelligence must understand what has effectively been transferred to the machine and what remains under human control. Governance therefore becomes essential.

The key question is no longer simply whether an organisation uses artificial intelligence, but how much authority it delegates to artificial systems, what controls exist around those systems and how their outputs are verified.

Security in the era of synthetic intelligence

The work also addresses the transformation of cybersecurity in an environment increasingly populated by synthetic intelligence. Generative AI changes both defensive and offensive capabilities.

Artificial intelligence can support security teams in detecting anomalies, analysing large datasets, identifying vulnerabilities and accelerating incident response. At the same time, the same technologies can support increasingly sophisticated phishing campaigns, automated social engineering, synthetic identities, deepfakes, disinformation operations and scalable manipulation.

Cybersecurity therefore increasingly extends beyond the protection of systems and networks. It also requires protecting the decision environment itself. When humans and organisations operate in information environments containing machine-generated identities, content, arguments and evidence, the ability to establish authenticity, provenance and trust becomes a fundamental security function.

AI governance and geopolitics

Transformation also examines how different political and regulatory environments are responding to artificial intelligence. The work includes European Union and United States case studies, together with chapters addressing AI governance, the geopolitics of artificial intelligence and the strategic implications of access to computing infrastructure, energy, data and critical technological resources.

Artificial intelligence is becoming an increasingly important component of national capability, economic competitiveness and geopolitical influence. The development of advanced AI systems therefore cannot be analysed independently from technological sovereignty, semiconductor supply chains, energy infrastructure, cyber resilience and strategic autonomy.

From prediction to control

Rather than attempting to predict a single technological future, Transformation explores multiple prospective scenarios. The objective is to understand what mechanisms may allow individuals, organisations and institutions to maintain meaningful control while benefiting from increasingly capable artificial systems.

The analysis therefore focuses on questions such as: How much cognitive activity should be delegated to machines? What capabilities must humans retain? How can AI-generated information be verified? How should organisations maintain accountability when decisions are increasingly AI-assisted? What infrastructures and resources are required to sustain global artificial intelligence development? And ultimately, how can technological progress remain compatible with human autonomy and the physical limits of the planet?

Continuing the ADA educational programme

Transformation continues the research and educational direction developed through the ADA programme – Analyse, Decide, Act. The programme focuses on strengthening the ability of individuals and organisations to operate effectively in increasingly complex digital and informational environments.

This direction previously produced the guide “Artificial Intelligence and the Manipulation of Human Perception”, which examined how artificial intelligence, synthetic content and automated persuasion mechanisms can influence perception, beliefs and human decision-making.

With Transformation, the analysis expands from the manipulation of information environments toward the broader transformation generated by artificial intelligence across individuals, organisations, infrastructure and society.

Access “Transformation”

The work is available in both English and Romanian, open access on Zenodo under CC BY-ND 4.0.

Topics: artificial intelligence · AI governance · cybersecurity · AI Act · NIS2 · human–AI interaction · cognitive security · geopolitics of AI · digital transformation · open access.

The work is also presented on CyberAID, the free education platform created by ProDefence, with a focus on what it means for cybersecurity education.

More of the department’s work is listed on the research and innovation page.

ProDefence has published the educational guide “Artificial Intelligence and the Manipulation of Human Perception” in both Romanian and English. The guide was developed under the ADA programme: Analyze, Decide, Act, initiated by ProDefence, and is free to read and download in PDF format.

Key points

  • The guide explains how artificial intelligence can shape perception, beliefs and decisions.
  • It covers algorithmic amplification, deepfakes, AI-driven persuasion and the psychological weaknesses these exploit.
  • Each topic ends with practical protection and response measures.
  • Both language versions are available below as free PDFs.

Why ADA was launched

Technology now evolves faster than society’s ability to assess risk, verify information and respond to digital threats. Manipulation is no longer limited to human persuasion: artificial intelligence adds scale, automation and precise targeting, amplifying influence operations, fraud, disinformation and social engineering, often faster than people and institutions can react.

ADA was created to close this gap by turning high-impact topics into clear guidance, realistic scenarios and practical prevention and response steps.

What the guide covers

  • Algorithmic amplification and attention manipulation, which shape beliefs and behaviour.
  • Synthetic media and deepfakes (video, image and voice) used for deception, reputational attacks and fraud.
  • Persuasion and automation powered by language models, which make social engineering and narrative manipulation scalable.
  • Behavioural and psychological vulnerabilities that modern digital channels can exploit.
  • Practical protection measures: verification habits, decision discipline and what to do when manipulation is suspected.

The aim is not only awareness but resilience: helping readers recognise warning signs, reduce impulsive reactions and adopt safer digital routines.

Download the guide

Select a cover to open the PDF: the English edition on the left, the Romanian edition on the right.

Authors and partners

The guide was written by Mircea Constantin Șcheau and Alexandru Ciprian Angheluș.

Its development was supported by Cloud Security Alliance Romania, a chapter of the Cloud Security Alliance, and by specialists from the Romanian Intelligence Service (SRI), National Cyberint Center, the Romanian National Cyber Security Directorate (DNSC) and the Ministry of Internal Affairs, through the Romanian Police.

From publication to impact

ADA is a growing programme, not a single release. Through CyberAID and the CysEdu platform, ProDefence and its partners continue to add guides, simulations and courses on real-world risks: phishing and fraud, manipulation, AI-driven threats, incident response behaviour and digital safety for different audiences.

Institutions, educators, professionals and community organisations are invited to use the guide, share it responsibly and take part in the wider effort to strengthen societal resilience.

Frequently asked questions

Is the ADA guide free?

Yes. Both the English and the Romanian editions can be read and downloaded free of charge as PDF files.

Who is the guide for?

For the general public, educators, professionals and institutions that want to recognise and resist AI-enabled manipulation, deepfakes and social engineering.

What does ADA stand for?

Analyze, Decide, Act: the three steps the programme teaches for responding to digital manipulation and threats.

Published in July 2025 as part of the ADA campaign. The platform announced at the time now operates as CysEdu.

Can artificial intelligence recognise a phishing message better than a person? In the test described below, it did: an AI-based system flagged the message as malicious within seconds and explained why. The conclusion of the ADA campaign is still the same, though: technology helps, but the final decision is made by the person who receives the message, and that depends on education.

Key points

  • We tested a real phishing message with the AI-based analysis platform developed by ProDefence.
  • The system classified the message as malicious and explained the manipulation techniques it used.
  • Attackers use artificial intelligence too, to write more convincing messages without mistakes.
  • Real protection combines automated filters with people trained to recognise warning signs.

What we tested

Have you ever received a message that “seems” real, but something does not add up? As part of the ADA: Analyze, Decide, Act campaign, we ran such a message through our AI-based analysis platform.

The result: the message was identified as malicious, with a detailed analysis in a few seconds. The system recognised the manipulation technique, showed how it played on the recipient’s emotions and gave clear explanations that an ordinary user can follow. The full test report is available below.

Why technology alone is not enough

The same technology that detects phishing is used to produce it. Language models let attackers write flawless messages, tailored to each victim and in any language. The classic warning sign, “it has spelling mistakes, so it must be fake”, no longer works.

An automated filter can stop part of the attacks, but not all of them. The message that gets through reaches a person, and that person needs to know what to check.

Warning signs that still apply

  • Urgency: “your account will be blocked in 24 hours”, “final notice”.
  • Emotion: fear, curiosity, greed or the wish to help.
  • Unusual requests: card details, SMS codes, passwords, payments to a new account.
  • Sender and link: the address does not match the organisation and the link points to a different domain.
  • Channel: banks and public institutions do not ask for confidential data by email, SMS or phone.

The ADA rule is simple: analyze the message, decide only after checking through an official channel, then act (report it, delete it or alert your bank).

Education and prevention: the ADA platform

The campaign prepared the launch of a cyber education and prevention platform for children, parents, teachers, employees, managers and seniors, built on the “Analyze, Decide, Act” concept. It includes:

  • real attack scenarios;
  • phishing simulations and interactive tests;
  • quick-response guides;
  • content adapted to each type of user;
  • an AI assistant that explains how to recognise, avoid and respond.

The project is carried out by ProDefence in partnership with CYSCOE, the Cyber Security Cluster of Excellence. For organisations, ProDefence also offers cyber security training and analysis of suspicious messages and files.

The test can be continued interactively on CysGame (in Romanian), where free phishing missions and quizzes put users in front of concrete situations. On the other side, the AI tools used by attackers, CysAI gathers threat research and risk profiles of AI agents.

Frequently asked questions

Can artificial intelligence detect phishing?

Yes, AI-based systems can recognise many phishing messages and explain the techniques used, but they do not stop all of them, especially highly targeted ones.

Who wins, AI or phishing?

Neither side for good: attackers use the same technology. What makes the difference is the combination of automated filters and trained users.

What should I do with a suspicious message?

Do not click links or reply. Check through the organisation’s official channel, then report the message; if you have already entered bank details, contact your bank immediately.

Event report, November 2024.

In November 2024, ProDefence CEO Alexandru Angheluș spoke at the Regional Conference of Intelligence Analysts in Iași about fake cryptocurrency trading platforms and the role of artificial intelligence in spreading false information. The conference was organised by the Territorial Intelligence Analysis Unit together with the Romanian General Police Inspectorate and the Central Intelligence Analysis Unit.

Key points

  • Information analysis connects fragmented data into usable intelligence in cybercrime, violent crime and financial fraud cases.
  • The presentation covered how fraudulent crypto platforms work and how disinformation spreads through digital channels.
  • Artificial intelligence plays a double role: a tool for attackers and a defence against fake news and scams.
  • Cybercrime and traditional crime increasingly overlap, so analysts need digital investigation skills.

Why information analysis matters

In criminal investigations, information analysis turns scattered data points (transactions, communications, accounts, devices) into a coherent picture. Whether the case involves cybercrime, homicide, robbery or financial fraud, analysts help identify suspects, understand criminal patterns and direct limited resources where they matter most.

The presentation: fake crypto platforms and fake news

The keynote was titled:

“Financial fraud: fake cryptocurrency trading platforms. Fake news: the role of artificial intelligence in disseminating false information.”

It covered three topics:

  • how fraudulent trading platforms attract and deceive users in the cryptocurrency ecosystem, from advertising to the moment withdrawals are blocked;
  • the techniques used to spread disinformation, especially through social media and messaging apps;
  • the double role of artificial intelligence, used by criminals to scale fraud and false content, and by defenders to detect it.

The main message was the growing convergence between cybercrime and traditional crime: almost every investigation now has a digital component, which requires stronger capabilities in both digital investigation and intelligence analysis.

Collaboration between institutions and the private sector

As a member of CYSCOE, the Cyber Security Cluster of Excellence, ProDefence contributes experience in threat intelligence, digital investigations and AI-assisted analysis. The conference confirmed the importance of inter-institutional collaboration and public-private partnerships: shared information, joint training and operational agility.

How ProDefence can help

ProDefence supports lawyers, companies and institutions with technical analysis in cybercrime cases and with analysis of electronic fraud and suspicious files. Related reading: Financial fraud and fake news: the role of artificial intelligence.

Event report, November 2024.

On 21–22 November 2024, the Maritime University of Constanța hosted the first edition of the BlackSea Maritime Cybersecurity conference, themed “Countering cybercrime threats in the current context of the conflicts in Ukraine and the Middle East”. ProDefence CEO Alexandru Angheluș, also President of CYSCOE, presented a real malware-driven financial fraud case and chaired a closed roundtable on cybercrime investigation.

Key points

  • The conference was organised by the Maritime Cybersecurity Centre of Excellence (MARCYSCOE) with major Romanian universities.
  • Topics: the legal and operational difficulties of investigating cybercrime during armed conflicts, including attacks on critical infrastructure.
  • ProDefence presented a real case of financial fraud through data theft with malware.
  • A closed roundtable discussed investigation methods, institutional cooperation and training.

The conference theme

The digital battlefield increasingly overlaps with conventional warfare. Attacks on ports, shipping companies and energy infrastructure in the Black Sea region raise a practical question: how can such incidents be investigated when attackers, servers and victims are spread across jurisdictions, some of them in conflict zones?

Speakers, from cybersecurity specialists and lawyers to researchers, stressed the need for harmonised legal frameworks that support cybercrime investigations, especially those targeting critical infrastructure, in the face of hybrid threats.

The ProDefence case study

Alexandru Angheluș presented a real cybercrime case involving financial fraud through data theft via malware. The presentation followed the incident step by step:

  • the attack vector and how the malware reached the victim;
  • the data exfiltration mechanism;
  • the investigative steps used to reconstruct and document the incident.

The case illustrated the type of work ProDefence carries out in cybercrime investigation support, malware and fraud analysis and incident response.

Closed roundtable

Alexandru Angheluș also chaired a closed roundtable with representatives of law enforcement and academia specialised in cybercrime. The discussion covered:

  • new investigation methods for cyber incidents;
  • cooperation models between institutions;
  • the need for stronger education and training in digital investigations and cyber law.

ProDefence and CYSCOE

ProDefence is a founding member of CYSCOE, the Cyber Security Cluster of Excellence, and contributes technical capabilities and strategic input to its mission. Participation in events such as BlackSea Maritime Cybersecurity supports cross-sector collaboration against modern cyber threats. Related: ProDefence at the MARCYSCOE maritime cybersecurity course.

Event report, November 2024.

Alexandru Angheluș, CEO of ProDefence and President of the Cyber Security Cluster of Excellence (CYSCOE), took part as a cybersecurity expert in a workshop organised by the Ministry of Economic Development and Digitalisation of the Republic of Moldova under the Moldova Cybersecurity Rapid Assistance initiative, supported by the e-Governance Academy (EGA). The workshop focused on Moldova’s national cyber resilience and the protection of its critical digital infrastructure.

Key points

  • The workshop brought together government officials, technical specialists and international advisers.
  • Discussions covered operational security frameworks, threat response and institutional preparedness.
  • Alexandru Angheluș participated as an expert affiliated with EGA; ProDefence and CYSCOE were not directly involved in organising the event.

The workshop

The event served as a working platform for aligning good practices in cybersecurity implementation and capacity development. For a country building its national cybersecurity framework, such workshops help translate strategy into concrete institutional measures: clear roles, response procedures and protection of essential services.

Workshop of the Ministry of Economic Development and Digitalisation of the Republic of Moldova on cybersecurity capacity building

Contribution

In his role as an EGA-affiliated expert, Alexandru Angheluș contributed to the discussions on:

  • operational security frameworks;
  • threat response strategies;
  • institutional preparedness for incidents affecting critical infrastructure.

The participation reflects the recognition of Romanian cybersecurity professionals at regional and international level.

“This initiative represents a meaningful step in building a secure digital future for Moldova. I was honoured to contribute alongside a team of forward-thinking professionals who are shaping the region’s cybersecurity landscape.” (Alexandru Angheluș)

Acknowledgements

Special thanks to Cătălina Plinschi, Secretary of State for Digitalisation, for her leadership, and to Mihai Lupașcu, Director of the Moldovan Cybersecurity Agency, for his commitment to national digital security. Thanks also to fellow experts Elsa Neeme, Epp Maaten and Vitalie Varaniță for their professionalism and collaboration.

Related

ProDefence supports public institutions and companies with cybersecurity services, including NIS2 compliance consulting for organisations in Romania.

Event report, November 2024.

At the Cybersecurity Workshop for Travel Agencies, held during the Romanian Tourism Fair at RomExpo, ProDefence represented the Cyber Security Cluster of Excellence (CYSCOE) with a real case: a travel agency defrauded after stealer malware harvested login credentials, allowing criminals to access internal systems and make unauthorised payments.

Key points

  • Stealer malware collects saved passwords and session data from infected computers.
  • In the case presented, stolen credentials gave attackers access to the agency’s systems and payment flows.
  • Travel agencies are exposed because they handle payments, personal data and many third-party platforms, often without a security strategy.
  • Basic cyber hygiene measures can be applied immediately and block most such attacks.

The case: credential theft and fraud

Alexandru Angheluș, CEO of ProDefence and President of CYSCOE, walked participants through the incident. The malware harvested login credentials for several platforms used by the agency. With them, the criminals entered the internal systems and carried out financial transactions without authorisation.

“The scenario revealed how cybercriminals use credential-stealing malware to infiltrate business systems and commit fraud. The case underlines the need for cybersecurity education and incident preparedness in the tourism industry.” (Alexandru Angheluș)

Topics covered

  • cyber risks for travel agencies without a dedicated security strategy;
  • cyber hygiene practices that businesses can apply immediately;
  • real attack scenarios and practical lessons for the industry.

What travel agencies can do now

  • Do not save passwords in the browser on computers used for payments; use a password manager.
  • Enable two-factor authentication on booking, banking and email platforms.
  • Separate the payment computer from the one used for browsing and downloads.
  • Verify any change of bank details by phone, using a known number.
  • Train staff to recognise phishing and suspicious attachments.

Speakers and partners

  • Prof. Gabriel Raicu and Prof. Remus Zăgan, Constanța Maritime University;
  • Alexandru Angheluș, CEO of ProDefence, representing CYSCOE;
  • Dr. Alexandru Bobe, CiTyINNOHub;
  • Fraga Țăriuc, President, and Orsolya Bakó, Women4Cyber Romania.

Special thanks to Mihaela Patentasu, Cristian Barhalescu and the National Association of Travel Agencies (ANAT) for facilitating the workshop.

ProDefence’s role

Through CYSCOE, ProDefence supports cybersecurity awareness and capacity building in exposed sectors. Companies affected by similar incidents can turn to ProDefence for malware and fraud analysis, cybercrime investigation support and cybersecurity training.

Skip to content