For organisations
Understand the intended scope of autonomous activity and the evidence available about what happened.
Start a conversationCapability is
not authority.
Autonomous action. Accountable authority.
Autonomous AI should be able to operate effectively within the authority it has been given. Increasing capability should not, by itself, confer greater authority.
EviState is developing ES Authority to provide an independently inspectable evidentiary basis around autonomous machine authority and consequential action.
For organisations, security specialists and insurance partners.
The idea at a glance
Make authority understandable.
The objective is to help organisations understand the authority given to an AI system, its actions, and departures from its intended scope.
- 01
Defined authority
What has the organisation authorised?
- 02
Autonomous operation
Useful activity within the intended scope.
- 03
Visible boundaries
Understand where activity goes beyond that scope.
- 04
Inspectable evidence
Review recorded authority, action and context.
Built for scrutiny
A clearer basis for assessment.
We are exploring the questions that matter to those deploying, evaluating and underwriting autonomous systems.
For security specialists
Challenge the product direction, identify limitations and shape meaningful independent evaluation.
For insurers
Explore whether evidence of machine authority and action could inform risk assessment and incident review.
An important boundary
Development,
not a guarantee.
ES Authority is under controlled development. Its effectiveness, integration requirements and potential underwriting value remain subjects for testing and independent evaluation.
It is intended to complement existing security and governance controls. This is not a claim to solve AI alignment or guarantee containment. Evidence does not itself confer insurance cover, regulatory approval or a guarantee of safety.
Help us challenge the proposition.
Developing, deploying, evaluating or underwriting autonomous AI? Let’s talk.