AI policies fitted to your sector. Ready for further configuration.
With AiTrust you do not start from scratch. The platform provides baseline AI policies for the supported sectors, together with an initial configuration of the matching control mechanisms. You get a basic framework for managing the use of AI, covering the rules for data access, use of knowledge, validation of responses, actions of AI digital workers and human oversight.
From a ready policy to evidence
- Step 1: Baseline policyYou start from a ready policy for your sector. It sets which actions are allowed, forbidden or require approval.
- Step 2: Configuration in AiTrustThe policy comes with a baseline control configuration that you fit to roles, data, AI digital workers, processes and permissions.
- Step 3: Control of the actionIn the configured processes the mechanism allows an operation, restricts it, holds it or requires a human decision.
- Step 4: EvidenceThe record shows the zone, the policy decision and the human decision, which helps to show how the rules were applied.
Policies are not just documents
AiTrust lets you turn policies into technical mechanisms that control AI actions in the configured processes, enforce set restrictions and record the results of each check. For each of the six areas, which the readiness assessment also checks, you see what the baseline policy sets and which mechanism enforces it.
AI
Acceptable use of AI and the register of use cases
What the baseline policy sets
Which tools and models are allowed and what they are used for.
AiTrust mechanism
A model assigned to an AI digital worker, a department and a stage. A new model goes through verification before use.
Data
Data classification and information protection
What the baseline policy sets
Which data may reach a model and in what form.
AiTrust mechanism
A check before sending to the model, masking of data in queries and document permissions.
Knowledge
Knowledge sources and reliability of information
What the baseline policy sets
Which sources AI uses and how an answer is assessed.
AiTrust mechanism
Selected sources of the organization and the sources used in the answer pointed out, to the extent supported by the configuration.
Decisions
Validation of results and approval rules
What the baseline policy sets
When an AI result needs a human review and approval.
AiTrust mechanism
Operations of specified classes require a human decision before execution.
Agents
Permissions and autonomy of AI digital workers
What the baseline policy sets
Which tools an AI digital worker uses and what it may do without a person’s consent.
AiTrust mechanism
An agent passport with an approval threshold, operation and cost limits and grants to tools. The AI digital worker acts on its basis, and the passport can be suspended or revoked.
People
Oversight, accountability and incident response
What the baseline policy sets
Who approves, who reacts and how decisions are documented.
AiTrust mechanism
A record of events and decisions in the controlled path. The organization sets the scope of the record and how long it is kept.
Policy management
What the baseline policy sets
Who changes the rules, who approves them, how exceptions are handled and when the rules are reviewed again.
AiTrust mechanism
The organization sets the people, exceptions and review dates. We help to describe them during deployment.
You set what each mechanism covers in the configuration for your processes.
Your organization. Your rules.
Baseline policies are the starting point of the deployment. You can review, edit, extend and configure them to fit the requirements of your organization.
- Ready in AiTrust
Baseline policies for the sector
Rules on data, models, access, validation, AI digital workers, oversight and events that take into account the nature of the sector, its typical processes and laws.
- Ready in AiTrust
Baseline control configuration
Each baseline policy comes with an initial configuration of the control mechanisms, so you do not start from an empty system.
- Set by the organization
Your organization’s rules
Your own roles, data classification, levels of autonomy, approval thresholds, exceptions and accountability. The organization approves the final shape of the policies.
You do not have to build the rules from the ground up. You start from a prepared baseline and fit it to your needs.
Examples of what the baseline policies take into account in sectors
- Banking and finance
- Confidentiality, ICT risk, third-party providers and oversight of decisions, and the DORA requirements where they apply.
- Healthcare
- Health data, confidentiality, access by permission and the safety of the clinical process.
- Law firms
- Professional secrecy, client data and the verification of legal bases and sources.
- Public administration
- Citizens’ rights, the legal basis of action, transparency and accountability for decisions.
- Industry and infrastructure
- Technical data, trade secrets, access of AI digital workers to operational systems, safety and continuity of processes.
If needed, you can review the policies with a lawyer, taking into account the laws that apply to your business, internal regulations and customer requirements.
Customer data in a request to AI
Policy
Customer contact details do not reach an external model without a check.
AiTrust mechanism
Before sending, AiTrust shows what the model would see. Detected identifiers are replaced with tags, and a person chooses whether to cancel, edit or send a restricted version.
Evidence
The AI send report next to the result records the zone, the decision and the number of redactions.
The run recorded in AiTrust on synthetic data shows how the interface behaves in this example and configuration. The scope of control in other processes depends on the deployment.
Scope of enforcement
AiTrust enforces the configured rules in the processes and channels its mechanisms cover. It does not automatically block an employee who uses a private account in another AI tool. That is what controls on the organization’s devices and network are for.
Start with the readiness assessment
The free readiness assessment in six situations shows in which areas you need policies and mechanisms most.
Ready policies for your sector. Configuration fitted to the process.
Control enforced by technology. Rules you can develop together with your organization. We will show the baseline policies and the AiTrust control mechanisms on the example of your process.
Baseline policies and the baseline control configuration are an organisational and technical minimum to start with. They do not automatically mean that every legal requirement of your sector is met. Compliance with the AI Act is assessed for a specific use.