Who Is Accountable When Autonomous Spacecraft Risk a Collision?

Key Takeaways

• Orbital safety depends on coordination and evidence as well as tracking accuracy.
• Planned servicing encounters require different safeguards from accidental close approaches.
• Autonomous operation increases the need for records that explain consequential decisions.

Autonomous Spacecraft Expose an Accountability Problem

Stanford Law School’s Space Law & Policy Lab published Governing Orbits on September 14, 2026, examining collision prevention and liability in increasingly autonomous space operations. Developed with The Aerospace Corporation, the study focuses on the United States’ civil-space governance framework and its interaction with industry practice.

The Governing Orbits study identifies a problem that extends beyond detecting nearby objects. Operators need to know how to coordinate, which conduct is expected, and how responsibility can be assessed if an encounter causes damage.

Autonomy makes these questions more demanding. Software can help a spacecraft assess an encounter or select a maneuver, but the existence of an automated decision does not explain whether the operator acted reasonably. That assessment depends on the information available and the system’s authorized behavior.

The study distinguishes intentional interactions from unintentional ones. An intentional interaction occurs when a spacecraft approaches another object for a planned purpose, such as servicing. An unintentional interaction is an unexpected close approach that may require avoidance.

This distinction creates a more useful basis for governance than treating every close encounter as the same event. Planned proximity operations can involve consent and detailed coordination. An unexpected encounter may leave limited time to establish contact or agree on a response.

The report also separates active spacecraft from inactive vehicles and other nonfunctional objects. A maneuverable satellite can participate in avoidance. An unresponsive object cannot, even if its ownership remains identifiable.

These categories influence the obligations that make sense. Requiring two active operators to exchange plans is different from requiring a servicing vehicle to approach an inactive target safely. A rule designed for one encounter type may be incomplete or inappropriate for another.

New Space Economy’s coverage of servicing and inspection connects the issue with an expanding set of commercial activities. Servicing depends on intentional proximity, so avoiding every close approach would prevent the service itself.

The governance task is to distinguish acceptable proximity from unmanaged risk. That requires information about intent and capability, together with procedures that remain usable when software performs part of the decision-making work.

Collision Warnings Do Not Automatically Produce Coordinated Action

A predicted close approach, often called a conjunction, is an assessment based on observations and estimated future motion. It is not a statement that a collision will certainly occur.

The quality of the prediction depends on tracking data and uncertainty. New observations can change the estimated encounter, and a planned maneuver can make an earlier prediction obsolete. Operators need enough information to interpret updates rather than reacting mechanically to each warning.

Coordination adds a human and institutional layer. Someone must receive the information, determine whether action is required, and contact the other operator when appropriate. A published email address is of limited value if it is not monitored during the relevant period.

The Stanford study proposes verified contact arrangements and records of attempted communications. Those are policy options in the report, not evidence of a universal existing requirement. Their rationale is that coordination failures can arise even when the technical warning is adequate.

An ephemeris provides predicted positions over time. Sharing ephemerides can improve understanding of intended motion, but the information needs to be timely and compatible with the receiving system. Old data can create a misleading picture of where a spacecraft will be.

Autonomous maneuvering makes that issue more immediate. If a spacecraft changes its path according to onboard criteria, other operators need a suitable way to understand the resulting motion. Disclosure of relevant behavior can support coordination without necessarily exposing every proprietary design detail.

The report identifies reporting standards as one possible response. A standard could define what information must be exchanged and how often it needs updating. The details would need to account for different spacecraft capabilities and mission conditions.

New Space Economy’s explanation of traffic coordination provides context for the infrastructure supporting these exchanges. Information services can assist operators, but providing a warning does not automatically determine which operator must maneuver.

A coordination system also needs to distinguish technical inability from noncooperation. A spacecraft may lack propulsion or be experiencing a failure. Rules that assume every object can perform the same response would misrepresent the operating environment and complicate later judgments about conduct.

Intentional Encounters Need Consent and Mission-Specific Safeguards

Satellite servicing changes the safety question from avoiding another object to approaching it under controlled conditions. The servicing vehicle may need to inspect or physically interact with a target, making proximity an intended part of the mission.

The study proposes a risk-tiered authorization framework for these encounters. Under that concept, obligations would increase as the operation becomes more consequential. A distant inspection and a docking attempt would not necessarily require identical procedures.

Such a framework could connect authorization with the actual activity. The relevant risks depend on the vehicles and the planned interaction, not simply on whether the mission is described as commercial or experimental.

Consent is a separate issue. A target’s inability to respond does not automatically grant permission to manipulate it. Ownership and jurisdiction remain relevant, and a removal or servicing mission may need arrangements involving more than one organization or state.

The report proposes standardized consent and ownership instruments for debris-removal activity. Templates could reduce repeated negotiation, but they would not eliminate the need to identify who can authorize the operation. A standardized form cannot resolve an underlying dispute over authority by itself.

Pre-approach notification is another proposed safeguard. Advance information can allow the target operator or other relevant parties to assess the encounter and prepare for it. The usefulness of notification depends on adequate timing and sufficient detail.

Financial responsibility also needs to reflect the mission. A servicing operation can create risks for a valuable target as well as the servicing spacecraft. The report considers mission-specific financial assurance as a way to address those consequences.

The commercial relationship can allocate some risks by contract. It cannot necessarily determine the rights of unrelated third parties affected by debris or a collision. Private agreements and public obligations operate at different levels.

The discussion of active debris removal shows why these questions affect practical demand. A technically feasible removal service still needs an authorized customer and a workable allocation of responsibility. Legal clarity can influence whether a mission can proceed even after the engineering is sufficiently developed.

Liability Requires Evidence About Conduct and Causation

International space law distinguishes different circumstances of damage. Under the Liability Convention, damage caused elsewhere than on Earth’s surface to another state’s space object is addressed through a fault-based standard in Article III.

That is different from the convention’s absolute-liability rule for damage on Earth’s surface or to aircraft in flight. It is also different from a complete private commercial insurance system. The convention establishes a state-centered international framework, with separate questions arising under domestic law and contracts.

Fault can be difficult to establish when expectations are unclear. A collision alone does not reveal whether an operator ignored reliable information or followed a reasonable procedure based on incomplete data. The assessment needs evidence about what happened before the impact.

Autonomous systems add records that may be relevant to that inquiry. The software configuration and its input data can affect why a maneuver occurred. A later reviewer may need to reconstruct which criteria the system applied at the time.

That reconstruction becomes harder if essential information is missing. Tracking observations may not fully describe the spacecraft’s internal state, and external observers may not know about a propulsion fault or an rejected command. The absence of records can prevent a confident account of causation.

The report discusses a proposed rebuttable presumption of liability for operators that fall below specified standards. This would be a policy choice, not a statement of existing universal law. Its purpose would be to change how uncertainty is handled when conduct fails to meet defined expectations.

Such a proposal raises design questions. The relevant standard must be clear, and an operator needs a meaningful opportunity to show why the apparent failure does not establish responsibility. Otherwise, the presumption could punish circumstances unrelated to the damage.

Domestic tort law presents another possible route for some disputes, as the report’s appendix explains. Jurisdiction and applicable law can affect whether a claim proceeds and how it is evaluated. A single general statement about orbital liability cannot capture every legal relationship.

The practical implication is that collision prevention and accountability share an information requirement. Records useful for coordinating a maneuver may later help explain it. Designing those records before an incident is more reliable than trying to reconstruct missing evidence afterward.

Autonomy Needs an Explicit Operating Envelope

Autonomy describes a degree of decision-making without immediate human intervention. It does not identify a single technical design, and it should not be treated as equivalent to the use of machine learning.

A spacecraft can follow automated rules whose behavior is closely specified. Another system may use more complex methods to assess conditions and choose among responses. Governance needs to consider what decisions the system can make and under which constraints.

An operating envelope defines those permitted conditions and actions. For collision avoidance, it can describe when automated behavior is allowed and when the system must seek another form of control. The exact requirements depend on the mission and the available communication.

The report identifies a lack of sufficiently developed guidance for autonomous operations as one governance gap. Its concern is not that software automatically removes human responsibility. It is that existing practices may provide too little clarity about acceptable design and deployment.

Verification needs to examine interactions between systems. An avoidance maneuver that appears reasonable for one spacecraft may change the encounter for another. Independent automated systems can respond to the same situation using different assumptions.

Information exchange can reduce that uncertainty, but it must remain timely. A plan shared before an autonomous maneuver may become outdated once the spacecraft acts. Coordination procedures need to account for changes that occur without a ground controller issuing each command.

The autonomous ground-segment discussion provides commercial context. Automation can reduce repetitive work, but it also changes which decisions operators need to monitor and how exceptions reach responsible personnel.

A safety assessment should include degraded operation. A sensor failure or loss of communication can place the spacecraft outside the conditions under which normal behavior was demonstrated. The design needs a defined response rather than an assumption that autonomy will compensate for every missing capability.

Records should preserve enough information to assess those decisions afterward. That does not require making every technical detail public. It requires an appropriate evidentiary path through which authorized reviewers can understand the system’s behavior when an incident or dispute makes that necessary.

Industry Standards Can Help but Cannot Settle Every Public Question

The Stanford report treats governance as a combination of law and industry practice. Standards can define technical expectations more quickly than a new treaty, and operators may adopt shared procedures because they improve coordination.

Their value depends on adoption and quality. A carefully designed standard used by only a few operators has limited reach. A widely used practice can also contain weaknesses if it developed around a narrow set of missions.

Voluntary standards may influence contracts and insurance arrangements. They can help organizations demonstrate how they manage risk, but adherence does not automatically resolve legal responsibility in every case. The applicable law and the facts of an incident remain relevant.

The report raises concern about informal practices becoming entrenched before their consequences receive adequate examination. Large operators can shape expectations through the systems they deploy. Smaller participants may then face a practical need to accommodate those systems.

That possibility supports broad participation in standards development. Operators with different mission types can identify assumptions that work for one class of spacecraft but fail for another. Public authorities also have interests that extend beyond the commercial parties writing an agreement.

The report’s proposal for an operator “good standing” assessment illustrates another possible mechanism. A reputation-based measure could reward cooperative behavior, but its design would need transparent criteria and a process for correcting errors. Otherwise, the assessment could become an opaque barrier to participation.

New Space Economy’s discussion of traffic-management difficulties connects technical coordination with institutional differences. A common data format does not establish agreement about enforcement or acceptable risk.

International participation adds further complexity. A domestic regulator can impose conditions within its authority, but it cannot independently create a complete global operating regime. Coordination across jurisdictions remains necessary where spacecraft belonging to different operators interact.

The distinction between useful standards and public authority should remain explicit. Standards can describe how to exchange information or perform an operation. Decisions about mandatory obligations and remedies require an appropriate legal basis.

Regulatory Change Must Be Separated From Policy Proposals

Governing Orbits was completed during an active period of regulatory change. Its editorial note records the Federal Communications Commission’s adoption of a space-modernization order on July 22, 2026, after much of the research and drafting had occurred.

The report describes the order’s new Part 100 framework and associated requirements. It also states that the new part had not yet entered into force when the report was published. That historical qualification matters when describing what the authors were evaluating.

Adoption and effective operation are different stages. A rule can be approved before all provisions become applicable. A publication discussing an adopted framework should not automatically be read as evidence that every obligation was already enforceable on its publication date.

The study also contains recommendations that go beyond the regulatory developments it describes. Its proposed coordination requirements and liability mechanisms need to be labeled as options. They should not be mixed with existing obligations in a single undifferentiated account.

The authors’ stakeholder engagement occurred before the July order was released. The report explicitly notes that the interviews do not necessarily reflect industry views on the adopted rules. That limits claims about whether consulted organizations supported or opposed the later framework.

This distinction is useful for evaluating policy research more broadly. A study can remain valuable as regulations change, but its descriptions of legal status need to be read against their date. Its analytical questions may outlast particular provisions.

The report’s United States focus also limits direct transfer to other jurisdictions. Its institutional recommendations respond to American authority and licensing arrangements. Other countries may face similar coordination problems without sharing the same legal structure.

New Space Economy’s coverage of the in-space economy provides a wider commercial setting for these questions. More kinds of activity create more kinds of interaction, making precise authorization and responsibility increasingly relevant.

The report’s strongest contribution is its organization of those interactions. By separating planned approaches from accidental encounters and treating autonomy as a complicating condition, it helps identify which obligation belongs to which problem. That is more useful than assuming one general rule can govern every orbital event.

Summary

Orbital accountability depends on being able to explain decisions before and after an encounter. Tracking information is necessary, but coordination procedures and usable records determine whether operators can act together and whether later reviewers can assess their conduct.

Governing Orbits offers policy options for that task. Its recommendations should remain distinct from binding law, and its United States focus should remain visible when considering international application.

The most practical lesson concerns evidence. A spacecraft operator’s safety architecture should include a way to preserve the information that explains consequential actions. As autonomy expands, the ability to reconstruct a decision may become as important to accountability as the ability to issue a command.

Appendix: Useful Books Available on Amazon

• Handbook of Space Law
• Advanced Introduction to Space Law
• Routledge Handbook of Space Law
• Space Debris: Models and Risk Analysis
• Spacecraft Dynamics and Control: An Introduction

Appendix: Top Questions Answered in This Article

What problem does Governing Orbits examine?

The study examines collision prevention and liability in increasingly complex orbital operations. It focuses on the United States’ civil-space framework and industry practice. Its analysis distinguishes planned interactions from accidental close approaches and considers how autonomous decisions complicate coordination and accountability.

What is a conjunction?

A conjunction is a predicted close approach between objects in space. It does not mean a collision will certainly occur. Operators assess the prediction and its uncertainty, which can change as new observations arrive or a spacecraft changes its planned motion.

Why distinguish planned and unplanned encounters?

A planned servicing approach can involve advance consent and detailed procedures. An unexpected close approach may require a response within limited time. Different encounter types create different coordination needs, so treating them identically can produce rules that are incomplete or impractical.

Does autonomy remove operator responsibility?

Autonomy does not independently remove the operator’s responsibilities. It changes how decisions are made and what evidence may be needed to assess them. The relevant questions include which actions were authorized and whether the system operated within the conditions for which it was evaluated.

What is an ephemeris?

An ephemeris describes predicted positions over time. Sharing it can help another operator understand intended spacecraft motion. The information must be timely and usable, because an outdated prediction may no longer represent the spacecraft after a maneuver or operational change.

Is liability for orbital damage always absolute?

The Liability Convention distinguishes damage on Earth’s surface from certain damage in space. Article III uses a fault-based standard for damage to another state’s space object away from Earth’s surface. Domestic law and contracts introduce additional questions that the international convention does not resolve alone.

Can inactive debris be removed without consent?

Inactivity does not automatically eliminate ownership or jurisdiction. A debris-removal mission may need authorization and appropriate agreements with relevant parties. The Stanford study proposes standardized consent instruments, but those proposals do not themselves create a general right to remove another party’s object.

Why preserve autonomous-system records?

Records can help explain what information the system received and why it selected an action. That evidence may support incident investigation or legal assessment. Without it, external tracking alone may be insufficient to reconstruct the spacecraft’s internal condition and decision process.

Are the report’s recommendations already law?

The report contains policy proposals as well as descriptions of regulatory developments. They should be read separately. A recommended liability presumption or coordination mechanism does not become binding simply because the study proposes it, and adopted regulations may have later effective dates.

Can industry standards solve the entire problem?

Standards can improve technical compatibility and clarify operating practices. Their reach depends on adoption, and they do not automatically establish public enforcement authority. Legal obligations and remedies still require an appropriate basis, with international coordination needed where different jurisdictions are involved.

Appendix: Glossary of Key Terms

Conjunction

A predicted close approach between objects in space. The prediction includes uncertainty and can change with new observations, so it provides information for an operator’s assessment rather than certainty that an impact will occur.

Ephemeris

A description of an object’s predicted positions at specified times. Operators use ephemerides to understand expected motion, but the information must be updated when maneuvers or improved observations materially change the prediction.

Proximity Operations

Spacecraft activities performed near another object for an intended purpose. They can precede inspection or physical contact, requiring procedures suited to the vehicles’ capabilities and the consequences of an unintended approach.

Rebuttable Presumption

A legal starting assumption that a party can challenge by presenting sufficient evidence. The report discusses such a mechanism as a policy option for liability, rather than describing it as an established universal rule for orbital collisions.