FHI MODEL IMPLEMENTATION GUIDELINES
- ncameron
- 2 days ago
- 14 min read
Introduction
These Model Implementation Guidelines are adopted alongside the Declaration on Foreseeable Harm to Innocents to assist States in developing domestic procedures that give effect to the existing IHL standards as articulated in the Declaration. The Guidelines are not binding but offer a framework that States may adopt or adapt according to their legal systems, military structures, and operational contexts.
It is axiomatic that FHI must not and does not make new law. These Guidelines operationalise what existing IHL already requires when properly interpreted in light of modern capabilities.
The Guidelines are organised into four parts corresponding to the procedural obligations in the Declaration:
Part A: Verification Procedures (supporting Article 4)
Part B: Precautionary Measures and Alternatives Assessment (supporting Article 5)
Part C: Documentation and Record-Keeping (supporting Article 9), including Human Shields Assessment (supporting Article 7)
Part D: Review, Investigation, and Accountability (supporting Articles 10–11)
States with established targeting procedures may find that existing systems already satisfy many of these requirements. The Guidelines are intended to identify best practices rather than impose a single model.
PART A — VERIFICATION PROCEDURES
A.1 Pre-Strike Verification Requirements
Before authorising any attack, commanders should ensure that verification measures appropriate to the circumstances have been undertaken. These may include:
(a) Intelligence fusion: Collation and analysis of intelligence from multiple sources, where available, including human intelligence, signals intelligence, imagery intelligence, and open-source information.
(b) Real-time surveillance: Where operationally feasible, use of ISR platforms to confirm target identity and assess civilian presence immediately prior to attack.
(c) Pattern-of-life analysis: Review of historical activity patterns at or near the target location to assess likelihood of civilian presence at the planned time of attack.
(d) Collateral damage estimation: Formal assessment of expected civilian casualties and damage to civilian objects, using established methodologies.
(e) Local knowledge consultation: Where available and secure, consultation with personnel possessing knowledge of local conditions, civilian activity patterns, or cultural factors relevant to the assessment.
(f) Environmental assessment: Consideration of factors such as time of day, day of week, weather conditions, and local events (markets, religious observances, school hours) that may affect civilian presence.
A.2 Verification Standards
The standard of verification required rises with:
• The sophistication of available ISR and intelligence capabilities
• The time available before the attack must be executed
• The expected severity of civilian harm if civilians are present
• The nature of the target (residential structures, densely populated areas, and locations known to shelter protected persons warrant heightened verification)
These verification standards have been demonstrated to be operationally achievable within advanced military systems. The US DoD Law of War Manual (2023) identifies examples of feasible precautionary measures including 'gathering more information, such as visual identification of the target through intelligence, surveillance and reconnaissance platforms,' 'consulting personnel with relevant expertise,' and 'reviewing previously approved targets at reasonable intervals.' The Guidelines build on this demonstrated feasibility by providing systematic structure and explicit capacity-indexing.
A.3 Resolution of Doubt
Where doubt exists as to whether a person is a civilian, that person shall be presumed to be a civilian.
This presumption reflects established best practice in advanced military legal frameworks. The US DoD Law of War Manual (2023) requires decision-makers to 'presume that persons or objects are protected from being made the object of attack unless the information available at the time indicates that the persons or objects are military objectives.' The Manual further requires that such determinations be 'consistent with the obligation to take feasible precautions to verify that the objects of attack are military objectives.' These provisions confirm that such standards are operationally achievable.
Where doubt exists as to whether civilians are present, and where time and circumstances permit additional verification, such verification should be undertaken before the attack proceeds.
Where doubt cannot be resolved and civilians may be present, the attack should not proceed unless narrowly defined emergency override conditions in Article 8 are satisfied.
A.4 Verification Failures
Failure to undertake feasible verification measures constitutes a serious violation under Article 11 of the Declaration and may give rise to administrative liability. Where verification failures contribute to civilian casualties that were foreseeable with reasonable diligence, criminal liability may attach.
A.5 Capacity-Indexed Verification Duties
Verification obligations scale with available ISR and targeting capabilities. Where high-resolution or persistent ISR is available, failure to consult or utilise these tools constitutes a verification failure. Units with limited capabilities remain bound by feasible measures, but cannot discount measures that are feasible for higher-capacity forces within the same chain of command.
PART B — OPERATIONAL ISSUES, PRECAUTIONARY MEASURES AND ALTERNATIVES ASSESSMENT
B.1 The FHI Decision Sequence
Before any attack where civilians may be at risk, commanders should apply the following decision sequence:

Step 1 — Status Determination Is the person directly participating in hostilities?
YES → FHI does not apply; standard IHL targeting rules govern
NO or DOUBT → Proceed to Step 2 (doubt resolved in favour of civilian status)
Step 2 — Verification Has civilian presence been verified through all feasible measures?
Apply verification procedures in Part A
Verification duties are capacity-indexed: forces with advanced ISR bear heightened obligations
Step 3 — Foreseeability Assessment Are persons ascertainable as civilians foreseeably at risk?
NO → Attack may proceed (subject to standard IHL)
YES → Attack presumptively prohibited; proceed to Step 4
Avoidability is not a discretionary consideration but a gating determination: proportionality analysis is not reached unless foreseeable civilian harm has been shown to be genuinely unavoidable.
Step 4 — Avoidability Gate (Two-Stage Test)
Stage 1: Can any feasible alternative ELIMINATE the foreseeable harm?
YES → Attack should not proceed; employ the alternative
NO → Proceed to Stage 2
Stage 2: Can any feasible alternative SUBSTANTIALLY REDUCE the foreseeable harm?
YES → Attack should not proceed; employ the alternative
NO → Harm is genuinely unavoidable; proceed to Step 5
Step 5 — Risk Allocation Have combatants accepted risk rather than transferring it to civilians?
NO → Attack should not proceed
YES → Proceed to Step 6
Risk allocation at this stage is assessed as evidence bearing on whether foreseeable civilian harm was genuinely unavoidable, and does not operate as an autonomous prohibition independent of the avoidability determination.
Step 6 — Proportionality Is expected civilian harm excessive in relation to concrete and direct military advantage?
YES → Attack should not proceed
NO → Proceed to Step 7
Note: Foreseeable and avoidable harm is inherently excessive. Proportionality analysis applies only to the residual category of genuinely unavoidable harm.
Step 7 — Emergency Override (if applicable) Are ALL eight conditions of Article 8 satisfied?
NO → Attack should not proceed
YES → Attack may proceed; post-strike review automatically triggered
The eight cumulative conditions are:
Imminence - mass-casualty attack (tens, hundreds, or thousands of deaths) within minutes or hours, not days
No feasible alternative - strike is the only means of averting harm; all alternatives documented and rejected
Unavoidability confirmed - harm genuinely unavoidable through two-stage avoidability test
Decisive asymmetry - casualties from strike substantially lower by a very large margin than casualties prevented
Irreversibility - threatened harm cannot be mitigated by later action
Intelligence reliability - current, multi-source where feasible, highly reliable
Highest-level authorisation - pre-designated authority at highest practicable level
Contemporaneous documentation and automatic review - factual basis recorded at time of decision; independent post-strike review triggered.
This sequence must be applied in order. Avoidability must be exhausted before proportionality. The emergency override applies only in the narrowest circumstances. Documentation of each step is required under Part C.
B.2 Precautionary Measures Checklist
When ascertainable civilians facing foreseeable harm are present or likely to be present, the following precautions should be systematically considered:
Measure | Considerations |
Timing adjustment | Can attack occur when civilian presence is reduced? Night vs day, weekday vs weekend, outside market/school hours? |
Weapon selection | Can smaller munition, precision-guided weapon, or reduced yield achieve objective with less collateral effect? |
Approach modification | Can angle of attack, direction of approach, or standoff distance reduce civilian exposure? |
Target component | Can a specific component be targeted rather than entire structure? |
Warning | Can effective advance warning be given without compromising mission? Roof-knock, leaflet, broadcast? |
Delay | Can attack be delayed until civilians depart? Is target time-sensitive? |
Alternative means | Can objective be achieved through ground operation, capture, or non-kinetic means? |
Abort criteria | What civilian presence thresholds trigger abort? Are these communicated to operators? |
B.3 Alternatives Assessment
Before authorising an attack where persons ascertainable as civilians are foreseeably at risk, the authorising commander should be satisfied that:
(a) The person or persons at risk are not directly participating in hostilities, or where doubt exists, that doubt has been resolved in favour of civilian status
(b) Feasible alternatives have been identified and evaluated through the two-stage avoidability test
(c) The reasons for rejecting each alternative have been articulated and documented
(d) Risk has been properly allocated - combatants have accepted increased risk where doing so would significantly reduce civilian harm; and
(e) The selected course of action represents the option reasonably expected to cause the least civilian harm unless the Article 8 emergency override applies, in which case the expected civilian harm must remain substantially lower — by a very large margin — than the imminent mass-casualty harm the strike seeks to prevent.
B.4 Risk Reallocation
Where precautions can reduce civilian harm only by increasing risk to the attacking force, such reallocation of risk should be undertaken where:
(a) The reallocation is feasible (i.e., does not require operations expected to result in the near-certain death of attacking forces); and
(b) The reduction in civilian harm is significant.
Factors relevant to assessing "significance" include the number of civilians at risk, the severity of expected harm, and the vulnerability of those at risk (with particular weight given to the presence of children).
Operational convenience, efficiency, cost, or generalised force protection considerations do not justify the rejection of feasible precautions.
B.5 The Two-Stage Avoidability Test
When persons ascertainable as civilians are at risk, commanders must apply the two-stage avoidability test. Stage 1 assesses whether feasible alternatives could eliminate harm. Stage 2 assesses whether feasible alternatives could substantially reduce harm. Each stage requires explicit documentation of the analysis and its evidentiary basis, including ISR feeds, sensor data, pattern-of-life analysis, CDE modelling, and operational alternatives considered.
B.6 Escalation of Targeting Decisions
States should establish clear escalation thresholds. The following are offered as models:
Expected Civilian Harm | Minimum Authorisation Level | Additional Requirements |
0 expected casualties | Unit commander (tactical level) | Standard verification |
1–5 expected casualties | Battalion/equivalent commander | Legal review recommended |
6–15 expected casualties | Brigade/equivalent commander | Legal review required; alternatives documentation |
16–30 expected casualties | Division/theatre commander | Senior legal review; full CDE; alternatives exhaustion documented |
>30 expected casualties OR sensitive site | National command authority | Ministerial/Secretary-level approval; comprehensive documentation |
Emergency override invocation | Pre-designated highest authority | All 8 conditions documented; automatic post-strike review |
These thresholds are illustrative. States should calibrate them to their command structures and risk tolerances.
PART C — DOCUMENTATION AND RECORD-KEEPING
C.1 Documentation Requirements
For each attack where persons ascertainable as civilians were present or reasonably likely to be present, the following records should be created and retained:
(a) Pre-strike intelligence summary: The intelligence basis for the attack, including target identification, assessed military value, and any information regarding civilian presence.
(b) Collateral damage estimate: The formal CDE, if conducted, including methodology and conclusions.
(c) Verification record: A record of verification measures undertaken and their results.
(d) Alternatives assessment: A record of precautionary measures and alternatives considered, and reasons for their adoption or rejection.
(e) Risk-allocation determination: Where risk reallocation was considered, a record of the assessment.
(f) Legal review: A record of legal advice provided, if any, and its conclusions.
(g) Authorisation record: Identification of the commander who authorised the attack and the time of authorisation.
(h) ISR recordings: Where available, surveillance or sensor recordings from the pre-strike and strike periods.
(i) Post-strike assessment: Battle damage assessment and any information regarding civilian casualties.
(j) Avoidability assessment: Findings from Stage 1 and Stage 2 of the avoidability test.(k) Capacity assessment: Statement of actual ISR and precision capabilities available and used.
C.2 Retention Period
All documentation specified in C.1 should be retained for a minimum of ten years from the date of the attack.
C.3 Security and Redaction
Documentation may be classified and access-controlled as required by national security considerations. However:
(a) Classification does not authorise destruction of records within the retention period.
(b) Redaction of sources and methods is permitted when records are provided to investigative or oversight bodies, but information essential to assessing legal compliance must be preserved and made available.
C.4 Human Shields Assessment
Where intelligence indicates that civilians may be present as human shields, the targeting assessment shall include:
(a) Classification: Whether shields appear voluntary or involuntary, based on available intelligence;
(b) Attribution: Whether the defending party placed civilians there or prevented their departure;
(c) Enhanced precautions: What additional measures have been taken given known civilian presence—warnings, timing delays, weapon selection, approach modification;
(d) Alternative exhaustion: What alternatives were considered, over what period, and why they were rejected;
(e) Proportionality recalculation: Whether expected civilian casualties remain proportionate given the military advantage, notwithstanding that the shielding party bears primary moral responsibility;
(f) Documentation: A contemporaneous record of factors (a)–(e) sufficient to support post-strike review and responsibility attribution.
C.5 Adverse Inferences
Where documentation that should have been created is absent, or where documentation has been destroyed, tampered with, or manipulated, investigative and oversight bodies may draw adverse inferences regarding compliance.
PART D — REVIEW, INVESTIGATION, AND ACCOUNTABILITY
D.1 Post-Incident Review
Every incident in which civilian harm occurs or is alleged should be subject to prompt initial review to determine:
(a) Whether civilian harm occurred
(b) Whether the attack was conducted in accordance with applicable procedures and the Declaration
(c) Whether further investigation is warranted.
Initial reviews should be completed within a timeframe appropriate to operational circumstances, but ordinarily within 30 days of the incident or the receipt of a credible allegation.
Investigations must assess whether the two-stage avoidability test was applied and whether the commander properly discharged capacity-indexed verification and precautionary obligations.
D.2 Emergency Override Review
D.2.1 Mandatory Review
Every invocation of the emergency override under Article 8 of the Declaration shall trigger mandatory independent review, regardless of whether civilian harm resulted. This review is additional to, and does not replace, any post-incident review required under D.1.
D.2.2 Timing
Review shall be initiated within 72 hours of the strike where civilian casualties occurred, and within seven days in all other cases.
D.2.3 Independence
The reviewing authority shall be independent of the command chain that authorised the strike. Independence requires, at minimum, that the reviewing authority:
(a) was not involved in planning, approving, or executing the strike;
(b) does not report to the commander who authorised the override; and
(c) has authority to refer findings for criminal investigation under Article 11(1)(d) of the Declaration without requiring approval from the authorising chain of command.
D.2.4 Scope of Review
The review shall assess whether each of the eight cumulative conditions in Article 8(1) of the Declaration was satisfied at the time of the decision, based on contemporaneous documentation and any other available evidence. The review shall address:
(a) whether the threat was imminent within the meaning of Article 8(1)(a);
(b) whether the scale threshold was met;
(c) whether alternatives were genuinely exhausted;
(d) whether the two-stage avoidability test was properly applied;
(e) whether the asymmetry was decisive;
(f) whether the intelligence met the reliability standard;
(g) whether authorisation was at the appropriate level; and
(h) whether documentation was contemporaneous and complete.
D.2.5 Determinations
The reviewing authority shall make one of the following determinations:
(a) Override properly invoked: All conditions were satisfied on the evidence available at the time. No further action required, though lessons learned shall be recorded.
(b) Override improperly invoked—administrative failure: One or more conditions were not satisfied, but the failure resulted from inadequate procedures, insufficient training, or good-faith error in judgment under extreme time pressure. Referral for administrative review under D.5 Tier 2.
(c) Override improperly invoked—grave violation: The override was falsely invoked or abused within the meaning of Article 11(1)(d) of the Declaration, including cases where the commander knew or should have known that the conditions were not satisfied, or where the override was invoked for purposes excluded under Article 8(3). Referral for criminal investigation under D.5 Tier 1.
(d) Insufficient documentation: Documentation required under Article 8(1)(h) of the Declaration was not created or retained. Adverse inference under C.5 of these Guidelines; referral for administrative or criminal review as appropriate.
D.2.6 Communication of Findings
Findings of the review shall be:
(a) communicated to the commander who invoked the override;
(b) transmitted to the designated national oversight body;
(c) recorded in a central register of override invocations maintained by each endorsing State; and
(d) included in periodic reporting under Article 19(4) of the Declaration.
D.2.7 Transparency
Endorsing States shall publish aggregate data on override invocations, including the number of invocations, the outcomes of reviews, and any accountability measures taken. Individual operational details may be redacted for security, but the fact of invocation and the review outcome shall be disclosed.
D.2.8 Systemic Review
A pattern of override invocations within a single command, theatre, or conflict shall trigger systemic review to determine whether the override is being misapplied or whether operational circumstances require doctrinal reassessment, as contemplated by Article 8(4) of the Declaration. Three or more invocations within a six-month period by the same command, or five or more invocations within a single conflict, shall presumptively trigger systemic review.
D.2.9 Record Retention
All documentation relating to override invocations, including the contemporaneous record required by Article 8(1)(h) of the Declaration, review findings, and any subsequent accountability proceedings, shall be retained for a minimum of twenty years.
D.3 Investigation
Where initial review indicates potential violations, a formal investigation should be initiated. Investigations should be:
(a) Prompt: Initiated without undue delay
(b) Impartial: Conducted by personnel not involved in the attack under review
(c) Thorough: Examining all available evidence, including documentation, ISR recordings, and witness accounts
(d) Effective: Capable of identifying responsibility and recommending appropriate action.
D.4 Transparency
To the extent consistent with operational security and the protection of sources and methods:
(a) Findings of investigations should be made public
(b) Findings should be transmitted to domestic oversight bodies (parliamentary committees, inspectors general, or equivalent)
(c) Findings should be communicated to affected communities.
States should publish periodic aggregate data on civilian harm incidents, investigations conducted, and accountability measures taken.
D.5 Accountability Framework
TIER 1 — GRAVE VIOLATIONS (Criminal Responsibility) | TIER 2 — SERIOUS VIOLATIONS (Administrative Responsibility) |
• Attacks launched in clear violation where civilian presence was known • Knowingly misrepresenting intelligence • Reckless disregard / wilful blindness • False invocation or abuse of emergency override • Falsifying or destroying documentation • Command responsibility failures | • Failure to take feasible verification measures • Failure to take feasible precautions • Failure to escalate targeting decisions • Failure to create or retain documentation |
REFERRAL: Military or civilian prosecutors for war crimes charges | CONSEQUENCES: • Formal reprimand • Mandatory retraining • Suspension from targeting duties • Loss of promotion eligibility • Administrative separation |
Tier 1 — Grave Violations (Criminal Responsibility)
Grave violations include:
• Launching or authorising attacks in clear violation of the Declaration where civilian presence was known or reasonably ascertainable
• Knowingly misrepresenting intelligence to secure approval
• Reckless disregard amounting to wilful blindness
• False invocation or abuse of the emergency override
• Falsifying or destroying required documentation
• Command responsibility failures
Grave violations should be referred to military or civilian prosecutors for consideration of criminal charges under domestic law implementing war crimes obligations.
Tier 2 — Serious Violations (Administrative Responsibility)
Serious violations include:
• Failure to take feasible verification measures
• Failure to take feasible precautions
• Failure to escalate targeting decisions as required
• Failure to create or retain required documentation
Serious violations should be addressed through administrative proceedings. In such proceedings, the operator or commander may be required to demonstrate that due diligence was exercised.
Consequences for serious violations may include:
• Formal reprimand
• Mandatory retraining
• Suspension from targeting duties
• Loss of promotion eligibility
• Operational restrictions
• Administrative separation
D.6 Command Responsibility
Commanders bear responsibility for violations committed by subordinates where the commander:
(a) Knew that violations were occurring or were about to occur; or
(b) Should have known, based on information available to the commander, that violations were occurring or were about to occur; and
(c) Failed to take all feasible measures to prevent or suppress the violations or to submit the matter to competent authorities for investigation and prosecution.
D.7 Victim Notification and Remedies
Where investigations conclude that civilian harm occurred, States should:
(a) Notify affected individuals or families of the findings, to the extent possible
(b) Provide information on available remedies, including compensation procedures
(c) Offer condolence payments or humanitarian assistance regardless of whether a legal violation is established.
Condolence payments do not constitute admission of liability and do not discharge obligations of full reparation where violations are established.
Conclusion
These Guidelines are offered as a resource for States seeking to implement the Declaration on Foreseeable Harm to Innocents. They reflect best practices drawn from existing military doctrine, international standards, and the practical requirements of effective civilian protection.
States are encouraged to adapt these Guidelines to their specific legal and operational contexts. The objective is not uniformity of procedure but consistency of commitment: ensuring that the Declaration's principles are given practical effect in the conduct of hostilities.
Adopted alongside the Declaration on Foreseeable Harm to Innocents at [place], on [date].


Comments