The first question is not which obligations apply, but where AI is actually used in your organization. CRISAM® records AI functions in applications, IT systems and cloud services, classifies them and documents what follows from that.
Legal status as of 23 September 2026
| AI system | Role | Classification | Status |
|---|---|---|---|
| Applicant pre-screening in recruiting | Deployer | open | |
| Customer chat in the service portal | Provider | in progress | |
| Text generator in marketing | Deployer | in progress | |
| Spare parts demand forecast | Deployer | done |
The EU AI Act is Regulation (EU) 2024/1689 on artificial intelligence. It classifies AI systems by risk: unacceptable risk and therefore prohibited, high risk with extensive obligations, transparency risk with disclosure obligations, and minimal risk with no specific requirements. The regulation applies directly in all EU Member States and distinguishes between providers and deployers.
The timeline has been changed: Regulation (EU) 2026/1744 of 8 Jul 2026, the Digital Omnibus Regulation on AI, postpones the obligations for high-risk AI systems to 2 December 2027 and, for AI in regulated products, to 2 August 2028.
The Digital Omnibus Regulation on AI has postponed the dates for high-risk AI. The following dates reflect this amended version.
Definitions, AI literacy and the prohibited practices apply.
Obligations for general-purpose AI models, national authorities and governance.
Transparency obligations under Article 50, enforcement by the authorities begins.
New prohibitions and the end of the transition period for providers of synthetic content.
High-risk AI systems under Annex III are subject to the full requirements.
High-risk AI in regulated products under Annex I must meet all requirements.
First the risk class: the intended purpose of the system determines how much is required. Then the role: whether you develop the AI or use it leads to different obligations.
The regulation distinguishes between those who develop an AI system and place it on the market, and those who use it. Both roles carry obligations, but not the same ones. Many companies are both: deployers of purchased systems, providers of what they build into their own product.
Today, AI is embedded in many applications, IT systems and cloud services without having been identified as such at the time of purchase. Without a complete overview of the AI functions in use, every classification remains incomplete.
AI literacy has applied since 2 February 2025, regardless of risk class: anyone who deploys or provides AI systems ensures that the people dealing with them have sufficient knowledge.
Eight areas of obligation. Which of them apply to you depends on the risk class and your role.
| Obligation | Applies to | What is required | How to map it in CRISAM® |
|---|---|---|---|
| Record AI systems | all | Before any classification is possible, you need to know where AI is used. This includes AI functions in purchased applications and cloud services | Central documentation of all AI systems with a complete overview of the AI functions used in applications, IT systems and cloud services. |
| Classify by risk class | all | Determine for each system whether it is prohibited, considered high-risk, subject to transparency obligations or falls under minimal risk. The classification must be justified | The AI systems overview dashboard shows all recorded systems with classification, status, risk rating and responsibilities. |
| Refrain from prohibited practices | all | Practices with unacceptable risk are prohibited, including harmful manipulation, social scoring and the generation of non-consensual intimate content | Systems that fall into this category are flagged as such in the model. Evidence that they are not in use is part of the documentation. |
| Ensure AI literacy | all | The people dealing with AI systems must have sufficient knowledge. The obligation applies regardless of risk class | Responsibilities are stored for each system. Training evidence can be managed as a measure with an owner and a deadline. |
| Obligations for high-risk AI, provider side | Annexes I and III | Risk assessment and risk mitigation, high-quality training data, logging to ensure traceability of results, technical documentation, information for deployers, human oversight, as well as robustness, cybersecurity and accuracy | Risks, controls, measures and compliance requirements are mapped for each system in the context of the entire GRC platform. The detail view shows the risk assessment, measure status and documentation of the requirements. |
| Obligations for high-risk AI, deployer side | Annexes I and III | Anyone who uses a high-risk system without having developed it themselves has obligations of their own under the regulation | CRISAM® covers provider and deployer obligations and supports both organizations that develop AI systems and those that use AI. |
| Transparency obligations | Article 50 | People must be able to recognize that they are dealing with an AI system. For providers of synthetic content, the transition period ended on 2 December 2026 | For each system, the applicable disclosure obligation and how it was implemented are recorded. The status feeds into the AI systems report. |
| Evidence for supervisory authorities and customers | all | Compliance must be documented and demonstrated on request, to authorities as well as to business partners | The documentation on meeting regulatory requirements is structured, audit-proof and audit-ready. Management and compliance reports support audits and regulatory evidence. |
From the first inventory to the report that documents the classification of every system.
Each AI system is recorded with its own entry: where it runs, who is responsible for it, which role you hold and which data it processes.
AI functions are embedded in applications and cloud services without being named as such in procurement. It is worth checking the contract documents.
Provider or deployerAnyone who develops a system and places it on the market is a provider. Anyone who uses it is a deployer. Both roles carry obligations.
Link to data protectionIf the system processes personal data, it also belongs in the record of processing activities.
The classification determines the effort. It is justified and remains traceable, even if the system changes later.
| AI system | Department | Role | Justification | Status |
|---|---|---|---|---|
| Applicant pre-screening in recruiting | Human resources | Deployer | Selection of employees | |
| Credit scoring in sales | Finance | Deployer | Creditworthiness | |
| Access control with facial recognition | Facility | Deployer | Biometric identification | |
| Performance evaluation in field sales | Sales | Deployer | Evaluation of employees | |
| Safety assistant in the product | Development | Provider | Component of a regulated product |
For each classified system, you assess which requirement is met and how this can be evidenced. The status can be analyzed per system and overall.
| Requirement | AI system | Actual | Target | Severity |
|---|---|---|---|---|
| Logging of results | Applicant pre-screening | 1 | 4 | |
| Evidence of data quality | Credit scoring | 1 | 4 | |
| Human oversight defined | Access control with facial recognition | 2 | 4 | |
| Technical documentation complete | Safety assistant | 2 | 4 | |
| Notice of AI use | Customer chat in the service portal | 3 | 4 |
For each gap, you create a measure with an owner and a deadline. Progress flows back into the same analysis in which the gap became visible.
AI governance is not a separate topic. Each system is linked to the same objects as information security, data protection and risk management.
| AI system | Asset in ISMS | Processing in DPMS | Risk in register | Control in ICS |
|---|---|---|---|---|
| Applicant pre-screening in recruiting | ||||
| Credit scoring in sales | ||||
| Access control with facial recognition | ||||
| Customer chat in the service portal | ||||
| Text generator in marketing | ||||
| Spare parts demand forecast |
| AI system | Open item | Owner |
|---|---|---|
| Access control with facial recognition | No dedicated risk stored in the register | Risk management |
| Text generator in marketing | Processing activity not yet entered | Data protection |
The result is the AI systems report: what is in use, how it is classified, which requirements are met and what is still being implemented. The same basis serves audits and customer inquiries.
| Evidence | AI system | Progress | Owner | Status |
|---|---|---|---|---|
| Inventory of all AI systems | all | 14.09.2026 | IT governance | |
| Classification with justification | 38 of 45 | 14.09.2026 | IT governance | |
| AI policy with approval | all | 06.03.2026 | Management | |
| Evidence of data quality | Credit scoring | planned 03/2027 | Finance | |
| Logging concept | Applicant pre-screening | being revised | Human resources | |
| AI literacy training evidence | all | planned 12/2026 | Human resources |
AI policy, resources, decisions on the use of high-risk systems.
Inventory, classification, assessment of requirements, reporting.
Link to the record of processing activities and the ISMS, shared view of risk.
Information on the AI functions in use, implementation in day-to-day business, training.
CRISAM® does not look at AI in isolation, but as an integral part of your corporate governance. Risks, controls, measures and compliance requirements are mapped in the context of the entire GRC platform.
A complete overview of the AI functions used in applications, IT systems and cloud services, with classification, status, risk rating and responsibilities.
Direct link to the information security management system and data protection management, for an integrated governance setup instead of three separate repositories.
CRISAM® supports both organizations that develop AI systems and those that use AI. Many companies are both at once.
The documentation on meeting regulatory requirements is structured, audit-proof and audit-ready. Management and compliance reports support audits and regulatory evidence.
The EU AI Act is applicable law and sets out specific obligations depending on risk class and role. ISO/IEC 42001:2023 is a voluntary standard and specifies how an artificial intelligence management system is established, operated and improved. The standard does not replace the regulation, but it provides the structure in which its requirements can be organized.
| Criterion | EU AI Act | ISO/IEC 42001 |
|---|---|---|
| Type | Regulation (EU) 2024/1689, directly applicable law | ISO/IEC 42001:2023, first edition, published on 18 December 2023, voluntary standard |
| Subject matter | Obligations per AI system, depending on risk class and role | Requirements for an artificial intelligence management system |
| Binding nature | Mandatory, with enforcement by authorities | Voluntary, certifiable like other management system standards |
| Perspective | Starting from the individual system: classify, meet obligations, provide evidence | Starting from the organization: policy, roles, processes, improvement |
| Outcome | Ability to provide evidence to authorities and business partners | Certificate for the management system |
| Time constraints | Dates are fixed, staggered until 2 August 2028 | No statutory deadline, the driver comes from the market |
| Mapping in CRISAM® | AI systems, classification, requirements per system, AI systems report | The structure of a management system is the same as for ISMS and data protection and can be mapped in the same model |
The obligations under the EU AI Act apply to you regardless of whether you are certified to ISO/IEC 42001. Conversely, the standard gives the regulation's requirements an organizational framework: policy, roles, risk assessment, internal audit. If you already operate an ISMS in accordance with ISO/IEC 27001, you know this structure. You can find an overview of all frameworks covered by CRISAM® under Standards and frameworks.
The first step depends on where you stand today.
Then you start with the inventory. It records AI functions in applications, IT systems and cloud services, including role, purpose and owner. Without this overview, every classification remains incomplete.
Then the focus is on the justification. For each system, you record which risk class applies and why. The dashboard shows classification, status, risk rating and responsibilities in one place.
Then you add AI governance without introducing another system. AI systems are linked to the existing assets, processing activities and risks instead of being maintained alongside them.
For this you need the CRISAM® GRC Platform and the CRISAM® DPMS Module. In the demo, we discuss your starting point and show what you already have in place.
The EU AI Act is Regulation (EU) 2024/1689 on artificial intelligence. It classifies AI systems by risk: unacceptable risk and therefore prohibited, high risk with extensive obligations, transparency risk with disclosure obligations, and minimal risk with no specific requirements. The regulation applies directly in all EU Member States and distinguishes between providers and deployers.
Yes. Regulation (EU) 2026/1744 of 8 Jul 2026, the Digital Omnibus Regulation on AI, amends Regulation (EU) 2024/1689, among others, and entered into force on 27 Jul 2026. As a result, the obligations for stand-alone high-risk AI systems under Annex III apply from 2 December 2027, and for high-risk AI in regulated products under Annex I from 2 August 2028.
Definitions, AI literacy and the prohibited practices have applied since 2 February 2025. The obligations for general-purpose AI models and the governance requirements have applied since 2 August 2025. The transparency obligations under Article 50 and enforcement by the authorities have applied since 2 August 2026. On 2 December 2026, new prohibitions and the end of the transition period for providers of synthetic content are added. The high-risk obligations follow on 2 December 2027 and on 2 August 2028.
The EU AI Act distinguishes four classes. Unacceptable risk: practices that pose a clear threat to safety, livelihoods and rights. They are prohibited. High risk: systems that pose serious risks to health, safety or fundamental rights, such as biometric identification, applicant selection and creditworthiness assessment. Transparency risk: systems where people need to know that they are dealing with a machine. Minimal risk: the vast majority of systems, with no specific requirements.
The European Commission names seven areas: adequate risk assessment and risk mitigation systems, high-quality training data to avoid discriminatory outcomes, logging to ensure traceability of results, detailed documentation for assessing conformity, clear information for deployers, appropriate human oversight, and a high level of robustness, cybersecurity and accuracy.
The regulation distinguishes between those who develop an AI system and place it on the market, and those who use it. Both roles carry obligations, but not the same ones. Many companies are both: deployers of purchased systems, providers of what they build into their own product. CRISAM® covers provider and deployer obligations.
The EU AI Act is applicable law and sets out specific obligations depending on risk class and role. ISO/IEC 42001:2023 is a voluntary standard and specifies how an artificial intelligence management system is established, operated and improved. The standard does not replace the regulation, but it provides the structure in which its requirements can be organized.
With an inventory that records purchased AI functions as well as systems developed in-house. CRISAM® documents AI systems centrally and provides a complete overview of the AI functions used in applications, IT systems and cloud services. The AI systems overview dashboard shows all recorded systems with classification, status, risk rating and responsibilities.
The regulatory information on this page comes from the following sources.
The information reflects the status of the cited sources as of the review date stated above.
In the demo, we walk through your case: inventory of AI systems, classification with justification, requirements per system and the final report.
You are currently viewing a placeholder content from Facebook. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More InformationYou need to load content from reCAPTCHA to submit the form. Please note that doing so will share data with third-party providers.
More InformationYou are currently viewing a placeholder content from Hubspot Embedded Content. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More InformationYou are currently viewing a placeholder content from HubSpot. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More InformationYou are currently viewing a placeholder content from Hubspot Meetings. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More InformationYou are currently viewing a placeholder content from Instagram. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More InformationYou are currently viewing a placeholder content from X. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More Information