To invest in carbon credits at scale, companies need to know what a credible tonne looks like. Increasingly, that answer is being written in Bonn, where the Paris Agreement’s Article 6.4 mechanism is setting the rules for how carbon credits are generated, verified, and issued — rules that apply to countries and companies alike. It is becoming the reference architecture other frameworks borrow from, and the choices made there this year will shape what companies can buy, claim, and report.
Two developments this summer showcase how Article 6 is already influencing wider policies and standards:
- Under the revised EU Climate Law, high-integrity international credits may count toward the bloc’s 2040 target beginning in 2036, capped at 5 percent of 1990 net emissions (see Beyond’s comments on the international credits consultations in May). On July 17, the European Commission published its proposal to operationalize a portion of that 5 percent through a revision of the Emissions Trading System, with international credits bought centrally rather than by individual companies. The Commission has signaled that future eligibility criteria for the types of credits allowed under this system will build on Article 6.4’s forthcoming rules.
- Europe is also working to scale the use of removal credits generated within the EU through its own certification standard, the Carbon Removals and Carbon Farming Regulation (CRCF). Under the proposal, the Commission would auction extra allowances and use the proceeds to buy and retire CRCF-certified removal. Nature-based removal pathways can be certified under the CRCF, but the ETS will not accept them yet due to permanence concerns. Before nature-based storage can qualify, the ETS proposal’s preamble says that liability for reversals has to be carried through “enforceable financial, institutional, and contractual mechanisms, such as a permanence fund.” The Commission has given itself until the end of 2034 to decide whether those pathways can qualify. Article 6.4 is working through the same question now, eight years ahead of Brussels. The CRCF Regulation required the Commission to assess its alignment with Article 6 by July 31, 2026. That deadline has passed, and the Article 6.4 rules it would assess against are still being written.
- The EU is not alone. The SBTi’s final Corporate Net-Zero Standard (CNZS) V2.0 is also tracking Article 6.4. The November 2025 draft suggested removals used for neutralization would have to carry corresponding adjustments. The final standard dropped the requirement but kept the disclosure: companies must report whether their removals carry corresponding adjustments, though they are not blocked from using credits that do not.
- More is coming, though. SBTi has flagged an upcoming call for evidence on whether contractual or financial mechanisms can substitute for physical durability — the same barrier keeping nature-based removals out of the ETS. That is also the question the Article 6.4 Methodological Expert Panel (MEP) put out for public comment in June. Article 6.4 will likely get there first, and SBTi will be tracking closely.
Both the EU and SBTi are landing on the same question, which is why the rules that Article 6.4 writes on permanence matter well beyond the Paris Agreement Crediting Mechanism (PACM). Beyond’s position across every consultation has been the same: durability is a spectrum, permanence can be contracted as well as geologically assured, and PACM will scale faster if its rules reflect that. We made that case again this summer, when several Article 6.4 rules went out for public comment. Below is a summary of our areas of focus.
The consultation
At its fourteenth meeting at the end of June, the Article 6.4 MEP released two documents for public input: a draft Reversal Risk Assessment Tool (Annex 07) and a concept note on implementing paragraph 62 (reversal risk remediation) of the Removals Standard (Annex 08). Together, these determine how many credits from each project are withheld against the risk of reversal, and what happens when a reversal occurs. Beyond filed submissions on both in July, drawing on feedback from our corporate members and our Contracted Durability White Paper co-authored with RMI and AFF.
Our submissions argue for better inputs and a wider toolkit, not for a lower bar.
Getting the Risk Numbers Right (Annex 07)
The draft tool applies first to activities that reduce non-renewable biomass consumption (e.g., clean cookstoves), but it is modular. The choices made here will carry into forthcoming methodologies for forestry, biochar, and geologic storage. We raised four issues.
- Buffer contributions are cancelled rather than held. Credits contributed for reversal risk are cancelled at issuance, so they cannot compensate a reversal later. They work as a fixed deduction rather than an actuarial reserve, and because nothing is held, nobody can check whether the amount withheld was accurate.
Our Recommendation: Hold contributions as a reserve, subject to periodic reconciliation against observed reversals. A buffer that can be tested is a buffer that can be trusted.
- The science is being applied past its validated range. The tool’s default values apply a method drawn from a peer-reviewed study of the contiguous United States, calibrated on US forest inventory data. The tool extends that method to produce binding country-level percentages for tropical jurisdictions where it has not been tested, using datasets resampled to 8 km by 8 km, from a disturbance model that explains 43 percent of observed variance. The climate scenario selection rests on a companion preprint that has not been peer reviewed.
Our Recommendation: Treat defaults outside the validated region as provisional, disclose the validated domain and the main sources of uncertainty, and shorten the review cycle from five years to two so values keep pace with a fast-moving field.
- Strong stewardship earns no credit. The tool sets the reduction factor for natural risk mitigation to zero, on the ground that no jurisdiction-wide programs could be identified. Prescribed burning, fire-break management, and community fire management are established practices with measurable risk mitigation at the project level.
Our Recommendation: Allow independently verified mitigation measures to reduce the natural risk rating, as the tool already allows for human-induced risk.
- High default values stack into unusable results. Natural risk defaults already exceed 30 percent of credited carbon in several high-biomass countries. Add a human-induced default, allow no credit for mitigation, apply no upper bound, and the result can price an activity out for reasons traceable to the model rather than the risk.
Our Recommendation: Correct the numbers going in rather than capping the number coming out. A cap would hide an inflated score without correcting it, and the same score would reappear the next time the tool is used. Validating the regional defaults and giving credit for verified risk management would bring the rating closer to the risk a project actually carries. Should the Supervisory Body (SBM) conclude that a ceiling is still needed after those corrections, it should rest on a stated scientific basis.
Building the Remediation Stack (Annex 08)
The concept note evaluates insurance, guarantees, and a Monetary Permanence Reserve largely as substitutes for the buffer pool. A footnote acknowledges the closely related Permanence Trust, originated by the American Forest Foundation, and paragraph 75 leaves room for a reserve that actively manages risk rather than only holds funds. Our central recommendation is to reframe the question.
No single instrument covers every risk type across every time horizon. Buffer pool contributions handle expected, activity-level losses. Insurance and guarantees address correlated or large-scale events that could drain the pool itself. An endowed reserve carries liability after monitoring ends and the project developer has moved on. These are layers in a stack, and the question for each is what job it does and under what conditions. That is the logic behind contracted durability.
Four recommendations follow:
- Insurers need information, not control. The Article 6.4 Removals Standard reserves administration of the buffer pool account to the registry administrator. Read strictly, that could deny insurers the data they need to underwrite it. Underwriting requires disclosure about the account, not administrative rights over it.
- A burgeoning market is not a failed one. Reversal-risk insurance is thin under Article 6.4 because the mechanism has not yet defined the contract terms a product would be written against. Where a standard has defined them, products have followed, and Verra’s durability pilot has already approved insurance-based instruments. Markets cannot mature against undefined requirements. Define the terms, approve products as complements to the buffer pool, and let the pilots inform the pace.
- Judge coverage by performance, not contract length. The concept note would require a single policy running the full crediting and monitoring period, which for removals means 45 years plus whatever a methodology adds. No insurer writes that policy; coverage typically runs one to five years and renews. Continuous protection can instead come from rolling coverage, and the concept note already contains the pieces: annual proof of continued cover, with an automatic fallback contribution to the buffer pool if it lapses, and renewable terms during the post-crediting period. We would add one piece it lacks, which is collateral against renewal risk, so that a policy repricing sharply or not being offered at all does not leave the buffer exposed.
- Start designing the reserve now. A Monetary Permanence Reserve, as the MEP calls it, depends on collecting fees early and compounding them for decades, so its horizon shrinks with every year of delay. The Removals Standard directs the SBM to develop procedures for such a reserve. The concept note analyzes the instrument without recommending the work begin, and we recommend closing that gap now. Every year of delay is a year of compounding interest the permanence reserve never gets back — and a year closer to the tipping points it exists to insure against.
What we are watching
The MEP considers the public comments in September and sends recommendations to the Supervisory Body, which decides in October — the last window this year to adopt new rules before COP31. Watch what follows: the Reversal Risk Assessment Tool is expected to extend to forestry, geologic storage, and biochar, where these defaults will carry far more weight. Beyond will be tracking each of these milestones and engaging directly with our members, NGO partners, and the Supervisory Body on the path to COP31.