AxiSentinel extends continuous AI compliance evaluation to satellite operators, Earth observation data and analytics firms, launch providers, satellite communications networks and the ground segment: the AI that screens conjunctions and plans the avoidance manoeuvre, watches spacecraft health on board, classifies radar and optical imagery, and turns pixels into findings that governments, insurers, lenders and traders rely on, evaluated against the licensing, debris, cyber, export-control and AI rules of every jurisdiction the operator licenses in and sells to.
Satellite operators were early to hand decisions to software: when to manoeuvre, what the spacecraft is doing wrong, what changed on the ground below. For years the only rules that mattered were the licence and the frequency filing. Between 2024 and 2026 debris deadlines, ground-segment cyber law, export rules, the EU AI Act and a proposed EU Space Act all arrived at the mission-control console at once.
Flight-software, conjunction-screening and analytics vendors each report on their own tools. No one evaluates the whole stack, across every jurisdiction an operator licenses in and sells to, with evidence a regulator, an insurer or a buyer of analytics would accept.
Space AI sits under licensing and debris rules, cyber law, export controls and AI-specific law, in every jurisdiction an operator licenses in, downlinks to and sells data into. AxiSentinel tracks every layer and builds RegDef packages the moment a rule is stable enough to evaluate against.
Applies to satellite and Earth observation AI placed on the EU market or used in the EU. Systems used exclusively for military, defence or national security purposes are outside its scope; dual-use systems sold commercially are not.
Proposed by the Commission on 25 Jun 2025 and under negotiation in the Council and Parliament through 2026. As proposed, it would apply from 1 Jan 2030 to EU operators and to non-EU operators providing space data or services in the EU.
Space is a sector of high criticality under Annex I: operators of ground-based infrastructure that support the provision of space-based services are in scope, with national laws transposed from 17 Oct 2024.
Applies from 30 Dec 2026 to large and medium operators, and from 30 Jun 2027 to most micro and small enterprises, with plot-level geolocation at the core of due diligence.
The FCC licenses US satellites and grants market access to non-US systems serving the US.
Private remote sensing space systems need a Commerce Department licence under 15 CFR Part 960, as rewritten in a rule effective 20 Jul 2020.
Spacecraft, remote sensing technology and related technical data are among the most tightly controlled exports, and models and training data can be technical data too.
Launch and orbital activities carried out from the UK are licensed under the Space Industry Act 2018, with the UK CAA as regulator.
Japan licenses launches and satellite control under the Space Activities Act and regulates Earth observation under a separate remote sensing law, both from 2016.
The international layer every national licence draws on, part treaty, part standard, part voluntary commitment.
General-purpose AI risk-management frameworks that already apply to every AI component in a satellite, ground or analytics stack.
From a single Earth observation analytics product to a full constellation with its ground network, AxiSentinel evaluates space AI against whichever rule actually applies, wherever the spacecraft, the ground station and the data sit.
Nothing about AxiSentinel’s core architecture changes for space. What changes is which RegDef packages are switched on and what telemetry the agents capture, from a conjunction-screening service to an imagery analytics pipeline.
Agents capture conjunction-screening, flight-software, ground-segment and Earth observation analytics telemetry on the cadence you configure, Always On or scheduled, never limited to a periodic audit cycle.
Every manoeuvre decision, analytic output and ground-system event evaluated against whichever licensing, debris, cyber, export or AI rule set actually applies, per event.
Every audit record is linked to the one before it in a signed, tamper-evident evidence chain, the kind of record an operator needs to show how each avoidance decision was really made.
Captures the conjunction data, probability estimate, model version and approval behind every avoidance manoeuvre, and every decision not to manoeuvre.
Ongoing evaluation of change-detection, object-detection and radar classification models for error-rate skew across regions, seasons and sensors.
Tracks where imagery, training data and model weights came from and where they went, with screening records for ITAR, EAR and EU dual-use controls.
End-of-life plans, ephemeris sharing and disposal timelines tracked against the FCC five-year rule, ISO 24113 and any Zero Debris Charter commitments.
A compliance-state change on one satellite, ground station or analytics product propagates fleet-wide as it happens.
AXI-Node agents deploy alongside mission control, ground stations, cloud processing pipelines and in-region environments, or fully air-gapped through the .axibatch format for sensitive and export-controlled programmes.
Nothing becomes a compliance finding until a qualified AxiLayer AI auditor reviews and signs it, not for a collision-avoidance model and not for an analytic sold as evidence.
Detection coverage, incident-handling and supplier-risk evidence for ground stations and mission control, mapped to NIS2 and the cybersecurity chapter of the proposed EU Space Act.
The same architecture monitoring a bank’s credit model or a hospital’s diagnostic AI now monitors a conjunction-assessment model or an Earth observation change-detection pipeline. What changes is which RegDef packages are switched on.
Space runs on AI decisions made continuously, in orbit and on the ground, and the rules governing those decisions are arriving within the same few years. AxiSentinel’s architecture already does this for software AI; space is a new RegDef surface on the same platform, not a new product.
Whether it’s a single analytics product or a full constellation with its ground network, AxiSentinel evaluates it the same way it evaluates any AI system: on the cadence you configure, Always On or scheduled, with full evidence, and with a human signature before anything counts as a finding.
