Legal Briefing: Why Articles 15 and 18 Cannot Cure CEVAWG: The Case for Revision Before Ratification

In February 2025, the 38th Ordinary Session of the Assembly of the African Union adopted the Convention on Ending Violence Against Women and Girls (AU-CEVAWG). The Convention grew out of a political initiative launched in 2020, formally mandated by the AU Assembly in February 2023 and accelerated following the Assembly’s February 2024 direction that negotiations be concluded in time for adoption the following year.

Its adoption marks an important political moment. Violence against women and girls remains pervasive across the continent, in domestic settings, conflict and post-conflict contexts, workplaces, migration settings and, increasingly, digital spaces. The need for coordinated regional action is not in dispute. The narrower legal question addressed in this briefing is whether the Convention, in its present form, is capable of delivering that protection.

AU-CEVAWG was developed within an already substantial legal framework, including the African Charter, the Maputo Protocol, the African Charter on the Rights and Welfare of the Child and, where applicable, CEDAW. A new treaty therefore needed to add clear normative value, consolidating and strengthening existing standards, closing identifiable protection gaps, articulating States’ due-diligence obligations and improving implementation, monitoring and accountability. This was especially important because many continuing barriers arise not from a complete absence of legal standards, but from weak implementation, inadequate services and resources, limited data and research, institutional constraints and conflict.

Following publication, African women’s rights organisations, feminist legal scholars, practitioners, activists and human rights defenders undertook the detailed public scrutiny that the formal process had not adequately facilitated. At a Public Reading and Strategy Consultation held from 15 to 17 September 2025, participants examined the Convention article by article. The resulting legal analysis acknowledged areas of progress but found that the deficiencies were not limited to isolated drafting errors. It identified narrow or inconsistent definitions, fragmented rights and standards, objectives and principles not translated into enforceable duties, incomplete State obligations, weaknesses in monitoring and enforcement, and significant omissions concerning prevalent forms of violence.

These concerns have led the Pause for Purpose campaign to call for ratification to be paused while the instrument is independently reviewed and strengthened. On 25 July 2026, the campaign presented a memorandum to the Pan-African Parliament calling for balanced scrutiny, an independent legal review and a pause in ratification advocacy.

Member States now face a consequential decision. The AU Commission launched a continental roadmap for signature, ratification, domestication and implementation on 6 May 2026, and advocacy for rapid ratification is intensifying. With only one ratification, the Convention remains short of the 15 deposited instruments required by Article 17 for entry into force. The available pre-entry-into-force window therefore remains open.

The choice before Member States is not between ratifying AU-CEVAWG and doing nothing. States remain bound by their existing obligations to prevent violence, protect women and girls, hold perpetrators accountable and provide effective remedies. The real choice is between ratifying a text whose deficiencies have been documented or using the present window to revise and strengthen it before States consent to be bound.

2. Purpose and central proposition

This briefing addresses a specific argument advanced in support of the immediate ratification of AU-CEVAWG. The argument is no longer simply that the text is legally sound, but that States should ratify it despite its acknowledged deficiencies because Articles 15 and 18 provide adequate means of managing or correcting them.

The first limb of this argument relies on Article 15, the Convention’s safeguard clause, which provides:

Article 15: Safeguard clause

No provision in this Convention shall be interpreted as derogating from the principles and values contained in other relevant instruments for the realisation of ending all forms of violence against women and girls.
In the event of a contradiction between two or more provisions of this Convention, the interpretation that favours the realisation of ending all forms of violence against women and girls shall prevail.

On the strength of this provision, proponents contend that existing and higher human rights standards will remain protected from any regressive effect of the Convention.

The second limb relies on Article 18, which establishes the following amendment and revision procedure:

Article 18: Amendment and revision

Any State Party may submit proposals for the amendment or revision of this Convention.
Proposals for amendment or revision shall be submitted to the Chairperson of the African Union Commission, who shall transmit the proposals to States Parties within thirty (30) days of receipt thereof.
The Assembly, upon the advice of the African Commission on Human and Peoples’ Rights and the recommendation of the Executive Council, shall examine these proposals within a period of one (1) year following notification of States Parties, in accordance with paragraph 2 of this Article.
The Assembly shall adopt the amendments in accordance with its Rules of Procedure.
The amendments or revisions shall enter into force in accordance with Article 17 of this Convention.

Article 18 is accordingly presented as a mechanism through which the Convention’s deficiencies can be corrected after it enters into force. On this view, the defects are tolerable because Article 15 “contains” their potentially regressive effects, while Article 18 makes them only “temporary.”

The central proposition of this briefing is that neither provision offers an adequate legal cure. Article 15 is an interpretive safeguard, not a source of missing rights or obligations. It may prevent the Convention from being interpreted as derogating from stronger standards, but it cannot supply the rights, definitions, obligations and enforcement mechanisms that the text omits. Nor can it repair the Convention’s conceptual, normative or structural defects.

Article 18, in turn, establishes a future amendment procedure; it is not a substitute for adopting a coherent and legally sound instrument at the outset. It offers only the uncertain prospect of political correction after the defective text has become binding, through a consent-based process that cannot guarantee that the necessary amendments will be initiated, adopted, ratified or uniformly accepted.

Section 3 of this breifing establishes the analytical distinction on which this argument rests. Sections 4 and 5 apply that distinction to Articles 15 and 18 respectively, while Section 6 sets out the legally responsible alternative.

3. The distinction between interpretation and correction

Whether Articles 15 and 18 can address the deficiencies in CEVAWG depends first on the nature of the legal problem involved. Three distinct categories must be separated: a conflict between applicable standards, ambiguity in the meaning of a provision, and an omission or structural defect. These problems are legally different. Interpretation may assist with conflict and ambiguity, but it cannot supply substantive content that the Convention does not contain.

3.1 Conflict

In international treaty law, a conflict arises where two applicable rules point towards different legal outcomes in relation to the same issue. The International Law Commission (ILC) adopts a broad understanding of conflict as a situation in which two rules or principles suggest different ways of dealing with a legal problem.^[1] A conflict may therefore arise where a CEVAWG provision appears to provide a lower standard of protection than another applicable regional, international or domestic legal norm.

Article 15(1) may assist in such circumstances; it provides that no provision of CEVAWG may be interpreted as derogating from the principles and values contained in other relevant instruments concerned with ending violence against women and girls. Article 15(2) further provides that, where two or more provisions of CEVAWG contradict one another, the interpretation that favours ending all forms of violence against women and girls must prevail.^[2] The safeguard may therefore prevent CEVAWG from being interpreted in a manner that reduces protection already secured elsewhere or favour the more protective interpretation where provisions within the Convention conflict. However, it is important to note that this function remains interpretive.

Article 15 does not amend the provision that produces the conflict. Nor does it transfer the full content of an external legal standard into CEVAWG. The stronger right or obligation continues to derive its legal force from the external instrument, and its application will depend on whether that instrument binds the state and can be relied upon before the relevant institution. Article 15 may manage the consequences of a conflict in a particular case, but it does not correct the Convention’s internal wording or structure.

Article 15 does not itself provide a complete rule for resolving conflicts. It refers to the ‘principles and values’ contained in ‘other relevant instruments’, but does not identify which instruments qualify, whether the instrument must be legally binding on the State concerned, or whether the stronger external right or obligation must prevail directly. It also does not explain how an interpreter should proceed where several relevant instruments contain different standards. Article 15 therefore directs attention to external sources without establishing a clear hierarchy or method for resolving inconsistency. This uncertainty weakens the argument that Article 15 offers a reliable cure for lower or conflicting standards within CEVAWG.

3.2 Ambiguity

Ambiguity arises where a provision is reasonably capable of more than one meaning. It may concern the meaning of a term, the scope of a state obligation or the relationship between different provisions. Unlike an omission, an ambiguous provision contains legal language, but the meaning or application of that language requires clarification.

CEVAWG’s definition of ‘gender’ provides a concrete example. Article 1(e) defines gender as “the roles, duties and responsibilities that are culturally or socially ascribed to a particular gender”. The definition is circular because it uses the term “gender” to define gender roles. It also does not clarify whether the concept includes socially constructed relations, identities or the unequal power relations between women and men that the preamble identifies as a cause of violence. Another example is the Convention’s designation of ‘cyberspace’ as a sphere in which violence occurs, seemingly separate from ‘public’ and ‘private’ life, when cyberspace can be both. Therefore, Article 15 may support a rights-protective interpretation, but it cannot replace this circular definition with clearer substantive language.

Furthermore, the ordinary rules of treaty interpretation govern this process. Article 31 of the Vienna Convention on the Law of Treaties requires a treaty to be interpreted in good faith, according to the ordinary meaning of its terms, read in their context and in light of the treaty’s object and purpose. Article 32 allows recourse to supplementary means of interpretation where the meaning remains ambiguous or obscure, or where the interpretation produces a manifestly absurd or unreasonable result.^[3]

Article 15 may guide the interpretive process where a CEVAWG provision is reasonably capable of more than one meaning. It may support a stronger, more rights-protective interpretation, but it does not guarantee that outcome. Its application will depend both on the language of the Convention and on the approach, knowledge and understanding of the court, commission or public authority interpreting it. The same provision may therefore be interpreted progressively by one decision-maker and more narrowly by another.

In all cases, interpretation remains constrained by the text. A rights-protective interpretation must be one that the wording can reasonably support. Article 15 does not authorise an interpreting body to rewrite the Convention or introduce words, rights or obligations that the instrument does not contain.

The World Trade Organization Appellate Body expressed this limitation in India: Patent Protection for Pharmaceutical and Agricultural Chemical Products, stating that treaty interpretation does not permit words or concepts that are absent from the treaty to be inserted into it.^[4] Article 15 may therefore resolve genuine ambiguity between plausible interpretations, but it cannot replace inadequate wording with stronger language that the drafters did not adopt.

The amendment provisions contain a further example of ambiguity. Article 18(5) states that amendments or revisions enter into force in accordance with Article 17. Article 17, however, regulates the initial entry into force of the Convention after the deposit of instruments of ratification or accession by fifteen Member States. It does not explain whether an amendment requires fifteen further ratifications or acceptances, whether it binds all States Parties or only those that accept it, or how an existing State Party expresses its consent to be bound by the amendment. The reference to Article 17 therefore does not provide a complete rule for the entry into force of amendments. This uncertainty cannot be resolved merely by stating that amendment remains theoretically available.

3.3 Omission or structural defect

An omission exists where a treaty fails to include a necessary right, state obligation, definition, remedy or enforcement mechanism. A structural defect concerns the organisation and coherence of the Convention as a whole. It may arise where related duties are scattered across different provisions, rights are not connected clearly to corresponding state obligations, definitions are incomplete, or implementation and monitoring arrangements do not form a coherent framework.

A central omission is the absence of an express and integrated due-diligence obligation. CEVAWG does not clearly state that States must exercise due diligence to prevent, investigate, prosecute and punish violence, provide reparation, and respond to acts or omissions by both State and non-State actors. Elements of these duties appear separately in Articles 5, 7, 10, 11 and 12, but they are not brought together as a coherent standard of State responsibility.

By comparison, Article 4(2) of the Maputo Protocol connects prevention, punishment, eradication, rehabilitation and reparation within one substantive provision. The CEDAW Committee similarly describes due diligence as requiring States to prevent, investigate, prosecute, punish and provide reparation for violence committed by non-State actors. Article 15 may preserve these external obligations where they are applicable, but it cannot insert the missing framework into CEVAWG itself.

These deficiencies cannot be resolved through interpretation. An ambiguous provision provides language from which an interpreter may select or clarify a legally permissible meaning. An omission provides no such language. Where CEVAWG does not contain a particular right or obligation, there is nothing in the Convention for Article 15 to interpret.

The existence of a stronger rule in another instrument may protect a claimant where that instrument applies, but it does not insert the missing rule into CEVAWG. In other words, article 15 cannot establish a missing due-diligence framework, replace an inadequate definition, reorganise fragmented obligations, create an omitted enforcement mechanism or convert a general commitment into a precise legal duty. Doing so would amount to revising the Convention rather than interpreting it.

Article 18 operates differently, it permits a state party to propose an amendment or revision, after which the proposal must be transmitted to the states parties, considered by the AU Assembly and adopted in accordance with the Assembly’s Rules of Procedure. The amendment or revision must then enter into force under Article 17. Article 18 may therefore provide a legal route through which omissions or structural defects could be corrected, but it does not itself correct them. Under the general law of treaties, amendment requires agreement between the parties and is legally distinct from interpretation.

A further structural inconsistency concerns the protection of women and girls’ human rights defenders. Article 1(m) defines such a defender as a “national of a State Party”, while Article 11(2) requires States to protect women and girls’ human rights defenders and women’s organisations. The nationality requirement appears to exclude non-national defenders, including migrant, refugee, asylum-seeking, stateless women and those who undertake cross-border human rights work.

This is difficult to reconcile with Article 7(2)(d), which expressly requires special protection for stateless women and girls, asylum seekers and refugees. It also departs from the UN Declaration on Human Rights Defenders, which recognises that everyone has the right, individually and in association with others, to promote and protect human rights. Article 15 cannot interpret the express word “national” to include persons whom the definition excludes. Correcting the problem would require the nationality restriction to be removed or revised.

The distinction is therefore decisive, article 15 may assist where CEVAWG conflicts with an applicable external standard or where its provisions are genuinely ambiguous. It cannot create missing rights and duties or repair defects in the Convention’s structure. Article 18 could theoretically alter the text later, but only through a separate legal and political process that often takes decades. Interpretation may manage uncertainty within an existing text; correction changes the text itself.

4. Why Article 15 cannot cure CEVAWG

Article 15 is frequently presented as the provision that will protect against, or correct, CEVAWG’s acknowledged weaknesses. That claim overstates what a safeguard clause can legally accomplish. Article 15 may guide the interpretation of the Convention and help preserve stronger standards found elsewhere, but its operation is constrained by CEVAWG’s wording and depends on the approach of the institution applying it. It cannot rewrite the Convention, fill substantive gaps or repair defects in its structure and enforcement framework.

4.1 Article 15 is a rule of interpretation, not a legislative power

Article 15, entitled ‘safeguard clause’, also known as a saving clause, ensures that a human rights treaty does not weaken or replace stronger protections in other international agreements. It prevents the treaty from being used to lower existing human rights standards, justify rights abuse, or limit broader rights recognised under other laws.^[5] Article 15 may prevent CEVAWG from reducing protections under other instruments, but it does not fix weaknesses in the Convention’s substantive or enforcement framework.

Article 15(1) provides that those interpreting the treaty should not read the convention in a manner that derogates from the “principles and values” that are found in other instruments. Article 15(2) addresses internal contradictions by requiring the interpretation that best advances the elimination of violence against women and girls to prevail. However, neither provision confers adjudicative powers.

As discussed in section2, article 31(1) of the Vienna Convention on the Law of Treaties (VCLT)^[6] provides that a treaty must be interpreted in good faith according to the ordinary meaning of its terms, read in context and in light of the treaty’s object and purpose.^[7] Interpretation can clarify the meaning, scope, and relationship of existing provisions, but it cannot lawfully replace negotiated treaty language with a materially different legal regime.

A provision which preserves external standards does not improve the internal quality of the Convention itself. Article 15 may influence the meaning attributed to an existing provision, but it cannot supply the normative content that the drafters omitted.

4.2. Article 15 will not automatically prevail in disputes

It is important to note that Article 15 will not always come up on top in every dispute. The operation of the Article hinges on the legal and institutional context. In practice, the effect accorded to Article 15 is likely to vary depending on several factors. These include the nature of the court, commission, administrative body, or other forum hearing the matter; whether the State concerned is bound by the external legal instrument being relied upon; the status of the relevant rule within domestic law; and the State’s constitutional framework governing the incorporation of treaties.

The application of Article 15 may also be influenced by any rules addressing overlapping or competing legal norms, the quality and strength of the evidence and legal arguments presented to the decision-maker, and the interpretive approach adopted by the institution responsible for resolving the matter

4.3. Progressive interpretation is institution-dependent

The case of Media Rights Agenda v. Nigeria^[8] serves as a primary example of progressive reasoning within the African human-rights system, in this case, the African Commission rejected a narrow and literal interpretation of the African Charter in favour of an approach that gave practical effect to the rights it protects. It resisted a minimalist interpretation of retroactivity, affirmed the primacy of international human rights obligations over restrictive national decrees, and recognised that public figures must tolerate a higher degree of criticism to safeguard freedom of expression and open public debate.

The decision therefore demonstrates how an interpretive body can use a purposive approach to prevent domestic law from weakening regional human rights protections.; this was important because it effectively invalidated military “ouster clauses” and ensured that the fundamental rights of the African Charter remain enforceable and cannot be “set aside” or made “illusory” by domestic legislation.

However, the same institution that adopted this progressive approach has taken a significantly more restrictive position in its interpretation and application of Article 59(1) of the African Charter, which governs confidentiality in proceedings before the Commission. The Commission’s current practice extends confidentiality beyond what is necessary to protect pending proceedings, including in circumstances where decisions have already been adopted but have not yet been formally transmitted to the parties or authorised for publication.

This restrictive approach has serious implications for access to justice. It limits access to information about proceedings and decisions, weakens the principles of openness and public scrutiny associated with a fair hearing, and can leave parties waiting for long periods without access to outcomes that directly affect their rights. Delayed publication also postpones implementation, restricts public accountability and prevents advocates from using the Commission’s jurisprudence to support similar claims elsewhere.

The practice also restricts meaningful participation by interested third parties. Where information about pending cases, the issues under consideration or adopted decisions remains unavailable, potential amici curiae may be unable to identify proceedings in which their expertise would assist the Commission or to seek participation at the appropriate stage. Civil society organisations and affected communities are similarly constrained in monitoring cases, supporting complainants and advocating for implementation.

This contrast is significant for CEVAWG. Article 14 assigns a monitoring role to the African Commission without addressing the restrictive confidentiality practice associated with Article 59(1).

The contrast between Media Rights Agenda v. Nigeria and the Commission’s approach to Article 59(1) demonstrates the central difficulty with relying on interpretation as a safeguard; the same institution may interpret one provision progressively and another restrictively. Interpretive outcomes are not fixed. They may vary according to institutional capacity, political pressures, the composition of the deciding body, its understanding of the relevant rights, and the quality of the arguments and evidence presented.

4.4. A safeguard cannot protect what has not already been secured elsewhere

Article 15 preserves existing standards; it creates none. It functions as a protective or preservative rule rather than an independent source of substantive rights. Its effectiveness therefore depends on the existence of a stronger external standard that is both legally applicable to the State concerned and practically enforceable by the affected woman or girl.

The protective value of Article 15 is consequently weakest precisely where protection is needed most. It may offer little or no practical protection where:

The State is not party to the stronger instrument. Nine AU Member States have still not ratified the Maputo Protocol; the Central African Republic became only its forty-sixth State Party in August 2025, 22 years after its adoption;
Domestic law provides a lower standard, leaving no higher national protection for Article 15 to preserve;
The relevant right or obligation is not clearly or comprehensively recognised in another applicable legal source, as may be the case with femicide and technology-facilitated violence; or
Access to a regional or international enforcement mechanism is unavailable or substantially delayed. Direct access to the African Court by individuals and NGOs requires the State concerned to make a declaration under Article 34(6) of the Court Protocol. Only a small minority of States have done so, and five (Rwanda, Tanzania, Benin, Côte d’Ivoire and Tunisia) have subsequently withdrawn their declarations.

For a woman living in a State that is not party to the Maputo Protocol or CEDAW, whose domestic law provides inadequate protection and which does not permit direct individual access to the African Court, Article 15 may preserve nothing because no stronger, applicable and enforceable standard is available to her.

Article 15 can therefore help maintain higher protections where they already exist, but it cannot fill normative gaps where those protections have not been clearly established elsewhere. Its protective effect is distributed in inverse proportion to need; women in States with stronger laws and broader treaty commitments may benefit from it, while those living under weaker legal regimes receive the least protection. The result is uneven protection across States, the opposite of what a regional convention should achieve through a consistent and reliable continental standard.

4.5. Article 15 preserves fragmentation rather than resolving it

Article 15 may also contribute to the continued fragmentation of the protection of women’s rights. Rather than providing a comprehensive and self-contained statement of rights and obligations, it potentially creates a framework in which the Convention must be read alongside a range of other legal sources, including the African Charter on Human and Peoples’ Rights, the Maputo Protocol, CEDAW, and, in some cases, customary international law. As a result, the scope and content of the protections available to women and girls may depend on the interaction of multiple legal instruments rather than on the Convention itself.

The need to interpret treaties in light of other international human rights instruments is not unusual. Human rights treaties are often read as part of a broader and interconnected legal framework. The difficulty arises, however, when external instruments are relied upon not merely to contextualise or reinforce the Convention’s provisions, but to identify or supply obligations that the Convention itself should have clearly articulated. In such circumstances, the protection afforded to women and girls becomes dependent on the existence, accessibility, and interpretation of other legal sources, thereby undermining legal certainty and the Convention’s ability to establish a clear and uniform regional standard.

While Article 15 may guard against regressive interpretations and help preserve existing protections, it cannot compensate for deficiencies in the Convention itself by supplying the clarity, completeness, and coherence that effective treaty drafting requires. Its operation in any dispute is contingent on forum and argument, its protection evaporates exactly where none exists elsewhere, and its method entrenches the fragmentation it is offered as answering. The burden of the ‘ratify now’ argument therefore shifts entirely to article 18, to which the next section turns.

5. Why Article 18 cannot cure CEVAWG

The second limb of the ‘ratify now, fix later’ argument rests on article 18 of the Convention. The claim is that, because CEVAWG contains an amendment procedure, its acknowledged deficiencies can be corrected after entry into force, and that ratification should therefore proceed without delay.

This section demonstrates that article 18 cannot bear that weight. In the law of treaties, an amendment clause is an operational provision designed for the prospective adjustment of a sound instrument to changed circumstances; it is not, and has never operated as, a remedy for defects that are known, documented and structural before the instrument has entered into force. Worse, the general law of treaties and comparative practice demonstrate that consent-based amendment procedures of the kind established by CEVAWG are inherently uncertain; they are often protracted, may fail to secure the necessary political support, and where amendments are adopted, it will bind only those States that accept them. Reliance on Article 18 therefore risks producing divergent treaty obligations among States Parties, fragmenting the very regional standard that the Convention is intended to establish.

5.1. The Legal Character of article 18

As set out in Section 2, Article 18 establishes a multi-stage amendment procedure involving initiation by a State Party, consideration within the AU’s political and institutional structures, adoption by the Assembly, and a further round of State ratification or accession. Adoption by the Assembly does not, by itself, bring an amendment into force. Under Articles 17 and 18(5), an amendment becomes legally effective only 30 days after 15 Member States have deposited their instruments of ratification or accession.

Three features of this clause should be noted at the outset. First, it is a classical, consent-based amendment procedure; starting with a proposal, political adoption, and a fresh round of state-by-state ratification. It contains none of the devices, such as tacit acceptance, opt-out amendment, majority amendment binding on all parties, that modern treaty practice has developed precisely because classical procedures proved too slow and too fragile to keep treaties functional.^[9]

Second, the drafters’ conjunction of ‘amendment or revision’ obscures a distinction that the International Law Commission (ILC) regarded as fundamental; ‘amendment’ denotes the formal alteration of particular provisions, whereas ‘revision’ refers to ‘a general alteration affecting the treaty as a whole’ a general review of the entire instrument.^[10]

As section 5.5 below shows, what CEVAWG requires is revision in a comprehensive sense, and a procedure built for discrete textual amendments, operating after entry into force, is structurally unsuited to deliver it.

Finally, Article 18 must be read alongside the residual rules in Articles 39–41 of the Vienna Convention on the Law of Treaties (VCLT), which apply to matters not expressly governed by the Convention’s amendment procedure.^[11]. Those rules are founded on the principle of State consent: an amendment cannot bind a State Party unless that State has accepted it. As explained below, this protection of consent exposes the fundamental weakness of the “fix it later” argument, there is no assurance that future amendments will be universally accepted or that they will produce a uniform legal standard across all States Parties. That protective architecture, is exactly what makes ‘fix it later’ legally hollow.

5.2 Article 18 is a procedure, not a guaranteed remedy

An amendment clause merely allows amendments to be proposed; it does not require States to consider, negotiate or accept them. When drafting the VCLT, the ILC deliberately stated only that “a treaty may be amended.” This wording treats amendment as a possibility, not as a right that imposes corresponding obligations on other States.^[12]

Between the proposal of an amendment to CEVAWG and its entry into force lies a series of uncertain steps, none of which Article 18 can guarantee; that a State Party will initiate the process; that States Parties will agree on the nature and extent of the defects; that the Assembly, operating through consensus-driven decision-making, will adopt the necessary amendments rather than dilute or defer them; that at least 15 Member States will ratify each amendment; or that acceptance will be uniform rather than partial and staggered.

These contingencies are not speculative. The comparative record of classical, ratification-based amendment procedures, which is the very design article 18 adopts, is one of chronic failure and delay:

Not one of the amendments adopted to the 1960 SOLAS Convention between 1966 and 1973 ever entered into force, for want of sufficient ratifications; as the International Maritime Organisation’s former head of legal affairs observed, under the classical procedure ‘amendments would never have entered into force and, as a consequence, every State would have adopted its own national rules… inevitably leading to chaos’. It was this failure that drove the invention of the tacit acceptance procedure.^[13]
The regional nuclear-weapon-free-zone treaties that adopted a unanimity amendment rule have never been amended; the 1968 Nuclear Non-Proliferation Treaty, whose article VIII requires majorities that include every nuclear-weapon state party for both adoption and entry into force, has likewise never been formally amended.^[14]
Even where ratification thresholds are eventually met, the timescales are generational; the revised African Maritime Transport Charter, adopted at Kampala on 26 July 2010, reached its fifteenth ratification, which is the same threshold article 17 of CEVAWG sets, only on 14 August 2025, fifteen years later.^[15]
A recent empirical survey of 491 amendment procedures across 691 multilateral agreements concludes that amendment procedures are ‘highly protective of the capacity for each state to oppose modifications’ in most cases securing a de facto veto over amendments and that this protectiveness has not significantly changed over decades.^[16]

The prospect of future correction under Article 18 is therefore legally and politically uncertain at every stage. Comparative practice shows that amendment procedures of this kind rarely result in timely and comprehensive reform. Significantly, those relying on Article 18 have yet to identify which provisions they acknowledge as defective or explain how the amendment procedure would remedy those defects.

It is unreasonable to ask States to ratify now and “fix it later” without first providing a clear analysis of the Convention’s deficiencies. Without such an analysis, States cannot assess the seriousness of the defects or determine whether they are, in fact, capable of correction through subsequent amendments. A promise of future repair that identifies neither the defects nor the proposed remedies is not a legal safeguard; it is merely a slogan.

5.3. Entry into force would entrench the defective text

Reliance on Article 18 requires States first to accept the Convention as it stands. The sequence matters; ratification would convert acknowledged defects into binding law before any amendment process had even begun.

Comparative practice suggests that years, and potentially decades, may pass between a treaty’s entry into force and the effective amendment of its provisions. During that period, States may model domestic legislation and policy on CEVAWG’s weaker or ambiguous formulations. They may also treat the later Convention as the new regional benchmark, creating a risk of downward harmonisation of existing protection standards.

Meanwhile, Courts, national human rights institutions and public authorities would be required to apply vague definitions and incomplete obligations, without the benefit of publicly available travaux préparatoires to guide their interpretation. Institutions would also develop implementation systems around the Convention’s defective architecture.

For example, Article 14 assigns monitoring responsibility to the African Commission without addressing the restrictive confidentiality practice associated with Article 59(1) of the African Charter. The delays and lack of transparency that already affect the Commission’s handling of decisions concerning gender-based violence could therefore become embedded in CEVAWG’s enforcement framework from the outset.

Ultimately, women and girls would bear the cost of seeking clarity through litigation, often over many years, on provisions whose deficiencies are already known.

African experience confirms that treaty provisions are not simply ‘fixed’ later. Thirty-eight years after the adoption of the African Charter, the African Commission still considers itself locked in by article 59(1): ‘If our Charter says all actions shall remain confidential, who are we to go against the Charter? We are bound by what the Charter says.’^[17] Twenty-two years after the Maputo Protocol, the weaknesses of articles 6(d) and 7(d) on marriage registration and matrimonial property continue to be litigated rather than corrected.^[18]

Nor can interpretation or soft-law guidance fill the gap in the interim. Interpretation and amendment are legally distinct processes; interpretation may clarify, and sometimes develop and expand, the meaning of an existing provision, but only amendment can alter the treaty’s text. As Special Rapporteur Waldock cautioned, using “interpretation” to perform the function of amendment would “seem[] to throw overboard the essential concept of the integrity of the text of a multilateral treaty.”^[19]

General comments and interpretive guidelines can elaborate standards that already exist in a treaty; they cannot create rights or binding obligations that the text omits. Ratifying a Convention that regresses from the Maputo Protocol and then relying on subsequent soft law to restore it to the standard of an instrument that is already binding is a lopsided strategy.^[20]

The key point is that ratification does not preserve the possibility of correction, rather, it makes correction harder. Once the text is binding, any amendment would have to displace provisions that States have already accepted, institutions have already operationalised, and some governments may have come to prefer those provisions precisely because they impose weaker obligations.

5.4. Amendments may produce a fragmented treaty regime

Even if amendments were eventually adopted and ratified, Article 18 would not necessarily produce a single continental standard. The general rule of treaty law is clear: “[t]he amending agreement does not bind any State already a party to the treaty which does not become a party to the amending agreement.”^[21] States that accept an amendment would therefore be bound by the revised text, while those that do not would remain bound only by the original Convention.

Article 40(5) of the VCLT adds another layer of complexity. Unless the treaty or the States Parties provide otherwise, a State joining the Convention after an amendment has entered into force may be governed by two versions of the treaty; the amended text in its relations with States that accepted the amendment, and the original text in its relations with States that did not. The result would be overlapping obligations rather than a single, uniform continental standard.^[22]

Article 18 of CEVAWG does nothing to displace these rules, it simply routes amendments back through article 17’s fifteen-ratification threshold, and because the Convention was drafted without preparatory works, there is no interpretive record from which even the operation of its own amendment clause could be clarified with confidence.^[23]

The predictable result is what treaty scholarship calls the ‘molecularization’ of a treaty; ‘a string of consecutive amendments that are binding on different sets of contracting parties has the disadvantage of creating a kaleidoscopic legal landscape, instead of a single unified approach. This impacts negatively on legal certainty and on the presumed efficacy of the treaty in question.’^[24] In practical terms, different States would be bound by different versions of CEVAWG, making it difficult to determine which obligations apply between particular States and undermining the Convention’s ability to establish a clear and uniform continental standard.

The pattern is well documented in practice; the Stockholm Convention on Persistent Organic Pollutants has been amended repeatedly to add prohibited substances, yet Australia has accepted none of the twenty additions, and India and China only eight and eleven respectively, leaving the parties bound by ‘differentiated bundles of obligations split among subsets of member states, with some adhering to the modified text and others following the original’.^[25]

Treaty drafters who genuinely anticipate the need for future adjustment usually include mechanisms to prevent this fragmentation, through tacit-acceptance procedures, opt-out amendment of technical annexes, or clauses under which an amendment accepted by a defined majority binds all parties, as the African Nuclear-Weapon-Free Zone Treaty itself provides.^[26]

CEVAWG’s drafters chose none of these devices. A convention whose purpose is to establish a clear and common continental standard on violence against women and girls would thus generate the opposite; divergent obligations across the continent, uncertainty for survivors and practitioners about which version binds which state, and fragmentation layered on top of the instrument’s existing incoherence with the Maputo Protocol and the African Charter.

5.5 The defects require comprehensive revision

The ‘fix it later’ argument implicitly assumes that the Convention’s problems are a short list of discrete provisions that targeted amendments could address. The article-by-article legal analysis demonstrates otherwise; the deficiencies are conceptual, normative, textual and structural, and they run through the entire instrument.^[27]

Objectives (article 4) and guiding principles (article 6) sit inside the operative text rather than the preamble, producing provisions that read as commitments but create no binding obligations. The definitions in article 1 are defective: ‘cyberspace’ instead of technology-facilitated gender-based violence; femicide named but never operationalised; women human rights defenders defined by nationality, excluding refugees, migrants and cross-border defenders; ‘family’ deferred entirely to national law, re-opening the public/private divide that African women’s rights instruments had begun to dismantle.

The right to be free from violence (articles 2 and 3) is severed from its scope and from its grounding in equality and non-discrimination. Article 5 omits the established due diligence framework, to prevent, protect, prosecute, punish, provide remedies, leaving states without a coherent account of what compliance requires. The Convention is silent on female genital mutilation, child marriage and conflict-related sexual violence. And article 14 assigns monitoring to an over-stretched Commission, with no dedicated treaty body of the kind that has proven effective under the Istanbul and Belém do Pará systems for example.

Recalling the ILC’s distinction, defects of this kind call for revision ‘a general alteration affecting the treaty as a whole’ not ‘amendment’ of particular provisions.^[28] They require systematic redrafting; internal cross-checking of definitions, rights and obligations; harmonisation with the Maputo Protocol, the African Charter and international standards; reconsideration of the obligation structure; and a coherent implementation and monitoring architecture. A post-entry procedure built for discrete textual changes, each requiring fifteen fresh ratifications and each binding only its acceptors, is structurally incapable of delivering such a comprehensive reform.

5.6 African Union practice supports revision or replacement before entry into force

It is sometimes suggested that adoption is a point of no return, leaving article 18 as the only path. AU practice demonstrates the contrary; where an adopted instrument has proven structurally deficient, the Union has revised, recalled or replaced it, including before it became operational.^[29] Examples include;

The Protocol on the Court of Justice of the African Union (2003) entered into force in 2009, but the court was never constituted; the Assembly instead merged it into the African Court framework through the Protocol on the Statute of the African Court of Justice and Human Rights (the 2008 ‘Merger Protocol’), replacing the earlier treaty entirely.
The African Convention on the Conservation of Nature and Natural Resources (Algiers, 1968) was comprehensively revised at Maputo in July 2003 after its shortcomings became apparent; the revised Convention superseded the 1968 text upon its entry into force in 2016.
The African Maritime Transport Charter (1994) was revised at Kampala in 2010 during a prolonged ratification period, the revised Charter entering into force in August 2025.^[30]

The argument here must be precisely formulated. It is not that AU law prohibits textual amendment under article 18. It is that AU practice consistently addresses structural problems through revised or replacement instruments and, where substantive expansion was required, through protocols, as with the African Court Protocol and the Malabo Protocol.

Reliance on a future protocol, moreover, would itself require a further round of negotiation and ratification, reproducing the very uncertainty the ‘fix it later’ position claims to avoid. These precedents also carry a sobering lesson about timing; even successful revisions took thirteen to fifteen years to enter into force. In AU treaty practice, ‘later’ is measured in decades, decades during which, on the ‘ratify now’ approach, the defective text would be the governing law.

The window for revision remains open. According to the African Union Commission’s official status list dated 22 May 2026, eight States had signed the Convention, but none had deposited an instrument of ratification. The Gambia’s National Assembly subsequently ratified CEVAWG on 09 June 2026, and Rwanda became the ninth signatory on 28 July 2026.

Domestic approval, however, is not equivalent to the deposit of an instrument of ratification. Article 17 counts deposited instruments. The Convention therefore remains well short of the 15 deposits required for its entry into force.^[31]

In conclusion, Article 18 offers only the uncertain prospect of future correction. It sets no timetable and cannot guarantee that States will agree on, adopt or ratify the necessary amendments. Comparative practice shows that consent-based amendment procedures often fail or take many years to produce results.

Even if amendments were eventually adopted, they would bind only the States that accepted them. By then, the defective text would already have entered into force and begun shaping domestic legislation, judicial decisions and institutional practice.

A possible future amendment process is therefore no substitute for adopting a coherent and legally sound instrument from the outset.

6. The legally responsible alternative (pause, review, strengthen then ratify)

Declining to rest on articles 15 and 18 does not mean abandoning the Convention or the continental commitment it represents. It means sequencing that commitment responsibly: pause, review, strengthen, consult and then ratify an instrument that can actually deliver.

This is the course the #Pause_for_Purpose campaign has put before member states, the Pan-African Parliament and AU organs ‘pause, listen, strengthen, then ratify’ and its posture is constructive, not obstructive; the demand is not the rejection of the Convention but an instrument of juridical value.

It is also, as section 6.2 below shows, the only course that preserves what article 18 cannot; a single, coherent text binding all states parties alike. We therefore recommend that the member states through the African Union take the following seven steps.

Pause the ratification campaign

Member states should refrain from ratifying, and AU organs and partners, including the ACHPR and UN Women, should suspend active ratification advocacy, including under the continental Roadmap for Signature, Ratification, Domestication and Implementation launched by the AU Commission on 6 May 2026, pending the review below. With only one instrument of ratification deposited against article 17’s threshold of fifteen, a pause forecloses nothing; it preserves the pre-entry window in which revision is simplest and cheapest.

Conduct an independent, expert-led, article-by-article legal review

The review should be conducted by independent legal experts experienced in treaty-making and should assess the Convention provision by provision. Its central task follows from the distinction developed in section 3 of this briefing and grounded in the law of treaties; to determine which concerns can lawfully be addressed through interpretation, which may clarify the meaning of a text, and which require revision, because only a change to the text can supply what the text omits. The comprehensive legal analysis published in December 2025 provides a starting point; if proponents contend that the Convention can achieve its objectives in its current form, that analysis should be answered on the merits through rigorous public and peer review rather than by slogan.

Map CEVAWG against the African Charter, the Maputo Protocol and relevant international standards

No baseline assessment of existing protections was undertaken before drafting. The review must therefore establish that baseline, the African Charter, the Maputo Protocol, the African Children’s Charter, CEDAW and its General Recommendations 12, 19 and 35, and applicable UN standards, so that the Convention’s added value can be defined and non-regression guaranteed. The analysis of the UN Special Rapporteur on violence against women and girls, which converges with civil society’s concerns on definitional ambiguity, accountability and process, confirms that this mapping is necessary; at the same time, contested interpretive positions advanced in that report should not be imported wholesale into any revision.

Identify regressions, omissions, ambiguities and structural defects

The review should produce a public, consolidated catalogue distinguishing the three legally different problems set out in section 3, conflicts with stronger existing standards; ambiguities; and omissions or structural defects. This is the analysis that ‘fix it later’ proponents have never supplied, and it is the indispensable foundation for any honest ratification decision.

Adopt an AU Assembly resolution calling for revision of AU-CEVAWG

Member states should table the matter through the Specialised Technical Committee, the Permanent Representatives Committee and the Executive Council, with Assembly champions proposing a resolution that recalls the Convention for revision and mandates a revision process; This is the same institutional route through which the Union replaced the 2003 Court of Justice Protocol and revised the Conservation Convention and the Maritime Transport Charter; revision should be presented as continuity of AU practice, not obstruction.

Conduct genuine consultation with women’s rights organisations, survivors and specialist legal experts

The revision process must repair the participation deficit that produced the current text; of 1,500 civil society organisations surveyed across Africa, only 33 per cent knew the Convention was being drafted and 93 per cent were never consulted. This is not merely a political grievance; it is a legitimacy defect in the legal sense. Consultation must therefore be on concrete draft text, not abstract aspirations; must span all AU regions and languages; must include survivors, women human rights defenders and frontline service providers; and must be documented, so that the revised Convention enters into force with the travaux préparatoires that article 32 of the VCLT presupposes as a supplementary means of interpretation, and with the ownership and legitimacy the current text lacks.

Resume ratification only once the revised text is coherent and rights-compliant

When the revised Convention states clear rights, precise due-diligence obligations and a credible monitoring architecture and demonstrably strengthens rather than duplicates or dilutes the Maputo Protocol and the African Charter, ratification should proceed, with the full support of the movements now calling for a pause. AU practice shows that revision during a ratification window is institutionally accepted and need not derail eventual entry into force. The goal has never been delaying for its own sake; it is an instrument that is fit for purpose. Pause. Listen. Strengthen. Then ratify.

This sequence is the legally responsible alternative because it aligns the political commitment to ending violence against women and girls with the legal quality that commitment demands.

7. Conclusion

The urgency of ending violence against women and girls is not in question. What is at issue is whether AU-CEVAWG, in its present form, provides the clear, coherent and enforceable legal framework that this objective demands. The deficiencies identified are not limited to ambiguities that interpretation can resolve or isolated provisions that might later be amended. They include substantive omissions, fragmented obligations and structural weaknesses that affect the Convention as a whole.

Article 15 can preserve stronger standards where they already exist and apply, but it cannot create missing rights, obligations or enforcement mechanisms. Article 18 offers only an uncertain and potentially lengthy route to future amendment, after the present text has become binding, and may leave States bound by different versions of the Convention. Neither provision can bear the corrective weight placed upon it.

The legally responsible course is therefore to use the still-open pre-entry-into-force window to pause ratification, conduct an independent legal review and revise the Convention through a transparent and participatory process. Revision before ratification is not a rejection of the Convention’s purpose; it is necessary to fulfil it. Africa’s women and girls should not be asked to bear the risks of a defective instrument today in exchange for the uncertain promise that it may be repaired tomorrow.

^[1] International Law Commission, ‘Fragmentation of international law: Difficulties arising from the diversification and expansion of international law’ UN Doc A/CN.4/L.682 (April 2006).

^[2] African Union Convention on Ending Violence Against Women and Girls arts 15(1) (2).

^[3] Vienna Convention on the Law of Treaties (adopted 23 May 1969, entered into force 27 January 1980) 1155 UNTS 331 arts 31–32.

^[4] India: Patent Protection for Pharmaceutical and Agricultural Chemical Products Appellate Body Report WT/DS50/AB/R (adopted 16 January 1998) para 45.

^[5] BG Ramcharan, ‘The Drafting History of Article 6 of the International Covenant on Civil and Political Rights’ in BG Ramcharan (ed), The Right to Life in International Law (Martinus Nijhoff Publishers 1985) https://doi.org/10.1163/9789004482296_006 accessed 30 July 2026.

^[6] United Nations, Vienna Convention on the Law of Treaties, United Nations, Treaty Series, vol. 1155, p. 331, 23 May 1969.

^[7] International Law Commission, ‘Draft Articles on the Law of Treaties with Commentaries’ (1966) II Yearbook of the International Law Commission 187, 221.

^[8] Media Rights Agenda and Others v. Nigeria (2000) AHRLR 200 (ACHPR 1998).

[9]On the spectrum of amendment designs and the rise of tacit acceptance and related procedures, see M Fitzmaurice & P Merkouris Treaties in motion: The evolution of treaties from formation to termination (2020) 193-201; L Bélanger & J-F Morin ‘Treaty amendment procedures: A typology from a survey of multilateral environmental agreements’ (2024) 37 Leiden Journal of International Law 62 68-75.

[10]Yearbook of the International Law Commission (1966) vol II 232 para 3; United Nations Final clauses of multilateral treaties: Handbook (2003) 96 & 107. See also M Fitzmaurice & P Merkouris ‘Amendment and modification of non-proliferation treaties’ in DH Joyner & M Roscini (eds) Non-proliferation law as a special regime: A contribution to fragmentation theory in international law (2012) 17 20-21.

[11]Vienna Convention on the Law of Treaties, 1969, arts 39-41. On their consciously ‘residual character’, see Fitzmaurice & Merkouris, 186; Yearbook of the International Law Commission (1966) vol II 231-232.

[12]Fitzmaurice & Merkouris, 188; the proposal to impose even a good-faith obligation of ‘due consideration’ on receiving states was omitted from the final text of the VCLT: Fitzmaurice & Merkouris, 190.

[13]Fitzmaurice & Merkouris (n 3) 197-199, quoting D Lost-Sieminska ‘The International Maritime Organization’ in M Bowman & D Kritsiotis (eds) Conceptual and contextual perspectives on the modern law of treaties (2018) 907 917.

[14]Fitzmaurice & Merkouris ( Joyner & Roscini) 35-37.

[15]African Union ‘Revised African Maritime Transport Charter enters into force: A new era for Africa’s maritime transport, port and shipping industry’ press release, 28 August 2025.

[16]Bélanger & Morin, 64 & 79-85.

[17]Statement of the African Commission before the Pan-African Parliament, 5th ordinary session of the 6th Parliament; see also the Commission’s statements during the panel on art 59(1) at its 81st ordinary session (‘[i]f a particular provision… outlives its provision, is it the responsibility of the Commission to remedy that mischief? Or do you go back to… the right forum?’).

[18]Protocol to the African Charter on Human and Peoples’ Rights on the Rights of Women in Africa, 2003, arts 6(d) & 7(d); the marriage-registration requirement has been the subject of advisory-opinion proceedings before the African Court, and matrimonial-property division remains contested before domestic courts, including the recent FIDA decision of the Kenyan Court of Appeal.

[19]Fitzmaurice & Merkouris, 216 (‘[i]nterpretation may change the meaning of the treaty but amendment and modification may result in the changing of the text of the treaty’); Yearbook of the International Law Commission (1966) vol II 236 (Waldock). On the contested and time-contingent character of interpretive outcomes generally, see M Fitzmaurice & P Merkouris ‘Treaty interpretation and its rules: Of motion through time, “time-will” and “time-bubbles”’ in Treaties in motion (2020) 121.

[20]Notes of Dr Ruth Nekura on member state engagement, Kenya, 12-13 November 2025 (Parts 2-3).

[21]VCLT art 40(4). As Fitzmaurice & Merkouris put it, ‘[o]nly states that have accepted or ratified the amendment are bound. States that have refused to do so remain bound only by the original text of the treaty’: Fitzmaurice & Merkouris ( Joyner & Roscini) 37.

[22]VCLT art 40(5); Yearbook of the International Law Commission (1966) vol II 234 para 13.

[23]On the absence of travaux préparatoires and its interpretive consequences, see ISLA, Fòs Feminista & Akina Mama wa Afrika Legal analysis of the African Union Convention on Ending Violence Against Women and Girls (January 2026) Part 2.

[24]Fitzmaurice & Merkouris (Joyner & Roscini) 38.

[25]Bélanger & Morin, 63.

[26]African Nuclear-Weapon-Free Zone Treaty (Pelindaba Treaty), 1996, art 19(3) (majority ratification binds all parties); Chemical Weapons Convention, 1993, art XV(4)-(5) (tacit acceptance for technical annexes); Fitzmaurice & Merkouris, 193-201.

[27] Pause for purpose, Legal analysis, Part 1.

[28] Yearbook of the International Law Commission (1966) vol II 232 para 3; United Nations Final clauses of multilateral treaties: Handbook (2003) 96 & 107.

[29]Pause for Purpose, Legal analysis, Part 3 sec 4.3; Nekura notes, Part 5.

[30]African Union ‘Revised African Maritime Transport Charter enters into force: A new era for Africa’s maritime transport, port and shipping industry’ press release, 28 August 2025.

[31]African Union Commission ‘List of countries which have signed, ratified/acceded to the AU Convention on the Elimination of Violence Against Women and Girls’ status list, 22 May 2026 (signatures: Djibouti, The Gambia, Angola, DRC, Liberia, Burundi, Ghana, Equatorial Guinea; ratifications: none); ‘NA ratifies AU Convention on ending violence against women and girls’ The Voice (Banjul) 10 June 2026; ‘Rwanda strengthens Africa’s fight against gender-based violence by signing landmark AU Convention’ Top Africa News 28 July 2026.