NIS-2 Directive: cybersecurity obligations for companies. Do you need support with implementation?More about NIS-2

Service: AI compliance

AI Compliance Consulting and External AI OfficerUse AI, but safely.

We make your use of AI legally compliant under the EU AI Act: with a risk analysis of your AI systems, a clear AI policy and training for your team. On request we act as your external AI officer. For companies in Germany, Austria and Switzerland, on site or remote.

Regulated, trained, supportedA clear policy, a trained team and a dedicated contact for all questions about AI.

Basics

What does the EU AI Act mean for companies?

The EU AI Act, Regulation (EU) 2024/1689, does not only apply to companies that develop AI. It applies to everyone who uses AI professionally, including ChatGPT, Copilot or a chatbot on your website.

Which obligations you have depends on two questions: what role does your company play, and which risk class does the AI system fall into?

As of November 2026, including the Digital Omnibus

To the AI Act check
  • Provider or deployer?

    Providers develop an AI system or place it on the market under their own name. Deployers use AI systems under their own responsibility. Most small and mid-sized companies are deployers.

    Note: anyone who substantially modifies a high-risk system or offers it under their own name can become a provider and then takes on considerably more obligations.

  • The risk class determines the obligations

    The AI Act divides AI systems into four risk classes. The higher the risk to people, the stricter the rules.

    • Unacceptable riskSocial scoring, emotion recognition in the workplaceProhibited since February 2025
    • High riskApplicant selection, lending, assessment of employeesStrict obligations from December 2027
    • Limited riskChatbots, AI-generated text, images and videosTransparency obligations since August 2026
    • Minimal riskSpam filters, translation, text suggestionsNo specific obligations, AI literacy still applies
  • AI literacy is mandatory, in a new version

    Since 2 February 2025, Art. 4 of the AI Act has applied to all providers and deployers, regardless of the risk class. Since 27 July 2026 it has been reworded by the Digital Omnibus: companies must take measures to promote the AI literacy of their staff. They do not have to guarantee a particular level of competence.

    If you document your policy and training, you can prove what you have done if it ever matters.

  • Data protection applies as well

    The AI Act does not replace the GDPR. As soon as personal data flows into an AI tool, you need a legal basis, usually a data processing agreement and, for transfers to third countries, additional safeguards.

    To data protection consulting

Our services

AI compliance consulting: our services

From the first assessment to ongoing support. We make sure your company can use AI without taking legal risks.

AI compliance under the EU AI Act

We map your AI systems to the risk classes of the AI Act and show which obligations follow for your company. Clear and without legalese.

What the AI Act requires

External AI officer

We take on the role of AI officer: as a dedicated contact for your team, for approving new tools and for all questions about AI.

More about the AI officer

Risk analysis of your AI systems

We record all AI tools in use and planned in an AI inventory and assess each one by risk class, obligations and required measures.

To the AI Act check

AI policy and training

A clear AI policy sets out what is allowed. Training brings the rules to your team and promotes AI literacy under Art. 4.

Policy and training

Sensible use of AI tools

Which tools suit your tasks, and where does AI bring real benefits? We assess use cases and set up an approval process for new tools.

More about tool use

AI security

We make shadow AI visible, prevent data leaking through prompts and check what AI tools in your company can access.

More about AI security

Automation with a sense of proportion

Recurring tasks in compliance and documentation can often be automated with AI. We show what makes sense and is legally sound. For extensive automation and data projects, our information management team is the right contact.

To information management

Core service

External AI officer: one dedicated contact for all AI questions

Is an AI officer mandatory?No. The AI Act does not require an AI officer. Even so, someone in the company has to be responsible for the AI inventory, policy and training. On request we take on this role externally.

You get a dedicated main contact with a deputy who knows your company. In person on site or remote, with regular reviews so that your AI compliance stays up to date.

Request an external AI officer

What your external AI officer takes care of

  • Build and maintain the AI inventory
  • Review and approve new AI tools
  • Maintain the AI policy and adapt it to new requirements
  • Organise AI literacy training
  • Point of contact for questions from the team
  • Report regularly to management

Internal or external AI officer compared

  • Expertise

    Internal: Has to be built up first

    External with machCon: Law, data protection, IT security and data science in one team

  • Costs

    Internal: Salary, training and time off from other duties

    External with machCon: Predictable by scope, without a full-time position

  • Independence

    Internal: Conflicts of interest with your own department possible

    External with machCon: Independent view from outside

  • Cover

    Internal: Often unclear during holidays or illness

    External with machCon: Dedicated main contact with a deputy

  • Staying current

    Internal: Track legal changes yourself

    External with machCon: We track changes such as the Digital Omnibus for you

  • Effort for you

    Internal: High

    External with machCon: Low

Risk analysis

Does the EU AI Act affect me? The AI Act check

Four questions give you an initial assessment. In the risk analysis we then record all AI tools in your company and assess them in detail.

What the risk analysis gives you

  • AI inventory of all tools in use and planned
  • Risk class for every AI system
  • Your role as provider or deployer
  • All obligations and deadlines that apply to you
  • Action plan, prioritised by urgency
AI Act check4 questions
Question 1: Do you use AI tools in your company, for example ChatGPT, Copilot or translation tools?

Then you are a deployer under the AI Act. Art. 4 on AI literacy applies to you, and an AI policy should set out which tools and data are allowed.

Check this carefully: in many companies AI is already being used without anyone knowing. This shadow AI makes an assessment worthwhile.

Question 2: Do you develop AI systems or offer AI under your own name?

Then provider obligations apply to you. Depending on the risk class, they range from transparency to technical documentation and conformity assessment.

Then you have no provider obligations. But make sure that modifications to purchased AI systems do not turn you into a provider.

Question 3: Do you use AI in HR matters, lending, education or to assess employees?

This may be a high-risk system. The obligations under Annex III apply from 2 December 2027. Preparation takes time, so it pays to start now.

Then your applications are probably not high-risk. Check this again for every new use case.

Question 4: Do customers interact directly with AI, for example via a chatbot or AI-generated content?

Then transparency obligations under Art. 50 have applied since 2 August 2026. People must be able to recognise that they are communicating with AI or seeing AI content.

Then the transparency obligations under Art. 50 probably do not affect you at present.

The check does not replace a full review. In a free initial consultation we discuss your result and the next steps.

All questions answered. Send us your answers right away, so you do not have to repeat them in the consultation.

Book a consultation directly

AI policy and training

AI policy and training: clear rules that reach your team

A policy only works if it is understood. That is why clear rules and training belong together.

What belongs in an AI policy

We draw up your AI policy to fit your tools and processes, short and easy to understand.

  1. Approved toolsA tool list showing which AI tools may be used for what
  2. Off-limits dataWhich data does not belong in AI tools, such as customer or HR data
  3. Approval processHow new tools are reviewed and approved
  4. Handling resultsCheck and take responsibility for AI results before passing them on
  5. LabellingWhen AI-generated content has to be labelled
  6. ResponsibilitiesWho is the point of contact and who decides

What our training covers

  • Safe use of AI tools in everyday work
  • The rules of your own AI policy
  • Data protection and confidential information
  • Limits of AI: errors, made-up facts and bias
  • The legal framework, explained in brief

For whom

  • Management
  • All employees
  • Power users

This is how you promote AI literacy under Art. 4 and can prove your measures.

AI security

AI security: use safely instead of banning

Bans do little. People who have no approved tools turn to private ones. We make sure AI is used safely in your company.

Shadow AI

Employees use AI tools that nobody has approved, often with private accounts. We make visible what is in use and move it to approved solutions.

Data leakage

What is typed into a prompt may leave the company. We clarify where data is processed and define what may go into which tools.

Access rights

AI assistants see everything the respective user has access to. We check permissions before a tool is rolled out.

Why permissions matter before the rolloutAn AI assistant such as Microsoft Copilot finds everything a user has access to. If folders are shared too generously, salary lists or contracts suddenly appear in answers. That is why we clean up permissions before the tool goes live.

Sensible use of AI tools

AI security does not mean less AI, but the right AI. We help you use tools in a targeted way.

  1. 1. Assess use casesWhere does AI bring real benefits, and where is the risk too high?
  2. 2. Select toolsSuitable tools with a sound contractual basis and data processing.
  3. 3. Regulate approvalA simple process through which new tools are reviewed and approved.

Deadlines

EU AI Act: these deadlines apply after the Digital Omnibus

With Regulation (EU) 2026/1744 the EU amended the AI Act in July 2026. The obligations for high-risk AI were postponed, many other obligations already apply.

As of November 2026, including the Digital Omnibus

  1. Applies2 February 2025Prohibited AI practices and AI literacy under Art. 4
  2. Applies2 August 2025Obligations for general-purpose AI models, supervision and penalties
  3. Applies27 July 2026Digital Omnibus in force, Art. 4 reworded
  4. Applies2 August 2026Transparency obligations under Art. 50, for example for chatbots and AI-generated content
  5. ComingDecember 2026Further prohibitions under Art. 5 and labelling of existing generative AI systems
  6. Coming2 December 2027High-risk AI under Annex III, for example in applicant selection and lending
  7. Coming2 August 2028High-risk AI in regulated products under Annex I, for example machinery and medical devices

What fines apply?

For prohibited AI practices up to EUR 35 million or 7 % of worldwide annual turnover, for breaches of other obligations up to EUR 15 million or 3 %. For small and medium-sized enterprises the lower amount applies in each case. The Regulation does not provide a separate fine for Art. 4 itself.

Does the AI Act also apply in Switzerland?

Yes, if you offer AI systems in the EU or the output of your AI is used in the EU. For Swiss companies with customers, subsidiaries or sites in the EU, the AI Act is therefore often relevant.

Our approach

Four steps to safe use of AI

We combine expertise in data protection, information security and data science so that AI benefits your company and risks remain manageable.

  1. 1. Assessment

    Which AI tools are in use, which are planned? We record everything in an AI inventory. It starts with a free initial consultation, on site with you or remote via video call.

    Free, 30 minutes
  2. 2. Risk assessment

    Classification under the EU AI Act, data protection and information security, per system and use case.

    On site or remote
  3. 3. Policy and training

    AI policy, responsibilities and AI literacy training for your team.

    Active implementation
  4. 4. Operation and support

    Review new tools, keep the policy up to date, regular reviews. On request as external AI officer.

    Ongoing support

On site or remote: your choice

We are happy to come to your company or work conveniently via video call, whichever you prefer. From Engen, Zug and Basel we can reach you particularly quickly. We support companies throughout Germany, Austria and Switzerland with the same care.

  • On site with youIn person at your company
  • Remote via video callFlexible and location-independent

Why machCon

AI compliance with experience

Data protection and IT security have been our daily business for years. Our AI compliance builds on this, with rules that work in practice and do not slow AI down.

Meet our team
machCon consultants working together at a screen

350+

companies supported

Pragmatic, not over-regulated

We implement what your company really needs, and no more. That way AI remains a tool for your team and does not become a never-ending project for administration.

ISO 9001 certified

Our internal processes are ISO 9001 certified. We attach great importance to consistently high consulting quality for you.

Dedicated contacts

Every client has a personal main contact and a permanent deputy, both of whom know your company. On site or remote, just as you prefer.

Here for you from Engen (Hegau), Zug and BaselIn person in the region and just as committed throughout Germany, Austria and SwitzerlandWhere we are on site

A large part of our team works in Engen in the Hegau region. From here we can quickly reach companies in Singen, Konstanz, Radolfzell, Stockach, Tuttlingen, Villingen-Schwenningen or Schaffhausen, and we are happy to come to you for audits, training and workshops.

In Zug and Basel we are right in the middle of one of Switzerland’s most important business regions. Here our consultants usually work directly at your company and are firmly integrated into your projects. Our clients come from a wide range of industries, from pharma and life sciences to local authorities. We support them in a wide variety of projects and provide ongoing support for data protection, IT security and quality management.

We support clients throughout Germany, Austria and Switzerland just as reliably, with dedicated contacts, remote meetings and on-site visits whenever they are needed.

Customer reviews

Become part of our success story

Companies achieving their goals in data protection, IT security and compliance with machCon.

“Working with machCon has taken us to a new level in data protection.”

“Their sound expertise and practical solutions helped us make our data protection policies not only legally compliant but also efficient. Thanks to their professional and proactive support, we no longer worry about fulfilling our data protection obligations in accordance with the law. A reliable partner we can recommend without reservation!”

allsafe GmbH & Co. KGClient, data protection
  • “Thanks to machCon’s NIS-2 and IT security consulting, we have significantly improved our security measures. Their expertise and customised solutions helped us meet NIS-2 requirements without difficulty. We particularly appreciate their straightforward and dedicated working style.”
    Fidel Dreher GmbHClient, IT security & compliance
  • “Fantastic business analysts! I am absolutely delighted to have machCon involved in my projects. They have an incredible drive and are solution-oriented while maintaining a friendly and positive working atmosphere.”
    Novartis Pharma AGClient, information management
  • “machCon is like Maverick from TopGun! I have to say I totally love working with machCon! They have very good project managers and coordinators as well as great business analysts and technical experts.”
    Roche Pharma Research ITClient, information management

Related services

AI compliance is part of the bigger picture

These topics complement your AI compliance and can be implemented with us from a single source.

Data protection

As soon as personal data flows into AI tools, the GDPR applies. Our external data protection officer is the ideal complement to AI compliance.

To data protection consulting

IT security

AI tools need secure access, clean permissions and well-thought-out emergency management. Our information security officer takes care of this.

To IT security consulting

Information management

For extensive automation and data projects: we classify information and make your data usable with business intelligence and data science.

To information management

Audit & assessment with ISO 42001

An AI management system under ISO/IEC 42001 is voluntary. It is worthwhile if AI is central to your business or customers ask for proof. We support you from the fit/gap analysis to certification.

To audit & assessment

Corporate governance

Clear rules for the use of AI belong in your compliance management. We anchor them in your governance, from the code of conduct to the whistleblower system.

To compliance and governance

FAQ

Questions and answers on AI compliance

The most important answers on the EU AI Act, the AI officer, AI literacy and the safe use of AI tools.

Your question is not listed?

In a free 30-⁠minute initial consultation we answer your questions personally, on site or remote.

Book initial consultation

Does the EU AI Act apply to my company?

Yes, as soon as your company uses AI professionally. The AI Act applies not only to providers but also to deployers, that is, companies that use AI systems. Which obligations apply in practice depends on the risk class of the system.

Our AI Act check gives you an initial assessment.

What applies if we only use ChatGPT or Copilot?

Then you are usually a deployer. Above all, Art. 4 on the AI literacy of your staff applies to you.

Data protection and information security come on top: which data may go into the tool, and where is it processed? An AI policy sets this out bindingly for your team.

Is an AI officer mandatory?

No, the AI Act does not require an AI officer. The role still makes sense, because someone has to be responsible for the AI inventory, policy and training.

On request we take it on as external AI officer.

What does AI literacy under Article 4 of the AI Act mean?

Companies that use AI must take measures to promote the AI literacy of their staff. The obligation has applied since 2 February 2025 and was reworded by the Digital Omnibus with effect from 27 July 2026.

You do not have to guarantee a particular level of competence. Training and a clear policy are the usual way to meet the obligation and prove it.

Do we have to label our chatbot?

As a rule, yes. Since 2 August 2026 the transparency obligations under Art. 50 have applied: people must be able to recognise that they are communicating with an AI, unless this is obvious.

Deepfakes, that is AI-generated or manipulated images, videos and audio recordings, must also be labelled. Exactly who bears which obligation depends on your role.

What is shadow AI?

Shadow AI means AI tools that employees use without the knowledge or approval of the company, often with private accounts. The risk: confidential data ends up in tools that nobody keeps track of.

We make shadow AI visible and move it to approved solutions. More under AI security.

May we enter personal data into AI tools?

That depends on the tool, the purpose and the contractual arrangements. We review the legal basis, data processing and third-country transfers and record the result in your AI policy.

What belongs in an AI policy?

Which AI tools may be used for what and which data does not belong in them. Plus responsibilities, approval processes for new tools, how to handle results and the labelling of AI-generated content.

The policy has to be easy to understand, otherwise it will not be followed.

Does the AI Act also apply to Swiss companies?

Yes, if you offer AI systems in the EU or the output of your AI is used in the EU. For companies with customers, subsidiaries or sites in the EU, the AI Act is therefore often relevant.

Do we need ISO 42001?

There is no obligation. ISO/IEC 42001 is a voluntary standard for an AI management system. It is worthwhile if AI is central to your business or customers require proof of responsible use of AI.

We support you with this through audit & assessment.

Use AI safely

Talk to us about your plans. In a free, non-binding 30-⁠minute initial consultation we clarify:

  • whether and how the EU AI Act affects your company
  • which AI tools you use and where the risks lie
  • which next steps make sense for you