The webinar was recorded on 11 June 2026
AI is already part of everyday work in many organisations. Employees use generative AI tools for everything from writing and ideation to analysis, summarisation and case preparation. This creates major opportunities for efficiency and better workflows.
But AI use also creates new risks if it is not governed.
In Promte's webinar about safe and compliant use of AI in the organisation, we focused on how organisations can give employees access to AI in a way that is useful, safe and aligned with compliance and governance requirements.
You can watch the webinar recording above if you want the full walkthrough.
When employees use AI tools without clear frameworks, the organisation can quickly lose control of which data is processed where, which tools are used and how AI-generated outputs are created.
In the webinar, we highlighted four risks organisations should pay attention to:
Legislation and compliance
AI use can conflict with GDPR, the AI Act or sector-specific legislation if data processing, documentation and accountability are not under control.
Unintended data sharing
Employees may enter personal data, confidential documents or business-critical information into open AI tools without the necessary agreements and security measures.
Vendor lock-in
If the organisation builds central workflows around one specific supplier or tool, it may later become difficult to change solution, move data or adjust processes.
Lack of governance
Without central login, user management, logging and clear guidelines, it becomes difficult to document AI use, quality-assure outputs and place responsibility if something goes wrong.
The point is not that organisations should stop using AI. On the contrary. They need frameworks that make it possible to use AI safely and responsibly.
A central theme in the webinar was shadow AI: employees using AI tools on their own without approval from the organisation.
This often happens for a simple reason: open tools work well, are easy to use and are quickly available. If the organisation does not provide a safe and usable alternative, employees find their own solutions.
That creates a challenge for the organisation. Even when the employee's intention is simply to solve a task more efficiently, the use may happen without control of data, compliance, documentation or security.
That is why blocking or banning is not enough. Organisations also need to give employees a real alternative: an approved AI platform that is safe enough for the organisation and good enough that employees actually want to use it.
There is no single solution that removes every AI risk. But organisations can significantly reduce the risk by combining technical, organisational and behavioural measures.
This includes:
Employees do not need to be experts in GDPR, the AI Act or every detail of sector-specific legislation. But the organisation needs to have made decisions and clearly communicated the applicable frameworks.
In the webinar, we presented Promte Guardrails, which is developed to help organisations reduce the risk of AI use.
Promte Guardrails can filter text and files before they are sent to an AI service. This could be when a user is about to enter CPR numbers, names, addresses, confidential information or other data that the organisation does not want processed in that tool.
An important point is that the filtering happens directly on the user's device in the browser. It does not happen by first sending the content to a large AI model to assess whether it may be processed. That matters because the risk may already have appeared if sensitive information is first sent to an external AI service.
With the browser extension, Promte Guardrails can also be used across other AI tools. This means the organisation can configure rules for when users should be warned, when an action should be blocked and which AI services may or may not be used.
Another important part of Promte Guardrails is the ability to get an aggregated overview of AI use in the organisation.
This gives the organisation insight into which types of AI services are being used and how often content is blocked or warned against. The purpose is not to monitor individual employees, but to give the organisation an overall picture of development.
That overview can be used constructively. If many employees try to use a tool the organisation does not want to allow, it may indicate a need for better information, clearer guidelines or a better approved alternative.
In this way, data can be used to improve AI governance without creating a monitoring culture.
Guardrails can reduce risk, but they do not solve the entire problem alone.
No filtering is 100 percent secure. Users may in some cases try to bypass rules, and external platforms can change their user interfaces, which means technical integrations need to be maintained. Guardrails should therefore be seen as one important layer in a broader effort.
A responsible AI strategy also requires training, clear policies, leadership anchoring and an approved AI platform that employees trust and want to use.
The most important point from the webinar is that security and usability must be considered together.
If the organisation's approved AI solution is too difficult to use or delivers too little quality, employees will tend to find other solutions. If the organisation provides a safe, compliant and usable alternative, it becomes easier to bring AI use into controlled frameworks.
AI is already changing how organisations work. The question is therefore not whether employees will use AI. The question is whether the organisation has created the right frameworks for that use to happen safely, responsibly and effectively.
Promte Guardrails is developed as a practical step in that direction: a way to create control without stopping AI use.
Organisations that want to test Promte Guardrails and the browser extension can contact Promte to hear more about the options and setup.
