Skip to content

What could change EPR next?

Tomorrow — 12 August 2026 — the great majority of the European Union's Packaging and Packaging Waste Regulation becomes generally applicable across twenty-seven member states. It does so with its flagship labelling implementing act unwritten, its harmonised label deadline effectively unset, and its producer-register format still pending. That combination is a fair emblem of this whole subject: the legal architecture of the next decade of packaging EPR is arriving faster than the technical content that would make it operate.

This article surveys what could actually change how EPR works over roughly the next ten years — digital product passports, item-level identification, automated sorting, chemical recycling, reuse mandates, recycled-content requirements and a global treaty. The organising discipline throughout is a four-way distinction that most commentary on this subject collapses, usually to the advantage of whoever is selling something:

  • In force — legally operative now
  • Legislated, not yet applicable — adopted, with a date
  • Proposed or consulted on — not yet law
  • Vendor claim or speculation — asserted, not established

Almost everything exciting in this field belongs to the last two categories. Saying so is the point of the article.

1. Digital product passports: correcting a widespread error

The most common mistake in commentary on this subject is to attribute packaging digital product passport obligations to the Ecodesign for Sustainable Products Regulation (Regulation (EU) 2024/1781). ESPR is the EU's general framework for setting ecodesign and information requirements, including digital product passports, through product-specific delegated acts. It is real, it is in force as a framework, and its central registry infrastructure went live in July 2026.

But packaging is not in ESPR's first working plan for 2025–2030 at all. The priority product groups are iron and steel, aluminium, textiles, furniture, tyres and mattresses, alongside energy-related products. Packaging is excluded precisely because it has its own dedicated instrument. Moreover, no ESPR delegated act had been adopted for any product group as of mid-2026; batteries carry passport obligations under a separate regulation from February 2027. The ESPR passport regime is therefore a live framework with an operational registry shell and essentially no substantive bite for any packaging producer — and none on its roadmap.

Packaging's digital obligations come instead from the PPWR (Regulation (EU) 2025/40), and they are narrower than "a digital product passport." The regulation requires harmonised material-composition and sortability labelling, and provides that where packaging participates in an EPR scheme, identification may be achieved through a QR code or other standardised, open digital-marking technology. Related provisions address marking of packaging containing substances of concern, and QR codes on reusable packaging communicating re-use system availability.

The important fact about all of this is its incompleteness. The European Commission confirmed in June 2026 that it will miss the deadline for the harmonised labelling implementing act, with a draft expected after the summer, followed by expert group discussion and public consultation. The downstream effect is that the harmonised material-composition label deadline — reported as twenty-four months after the implementing acts enter into force, or 12 August 2028, whichever is later — is now effectively unset, with 2028 a floor rather than a date. The producer-register format implementing act is likewise pending, delaying alignment with existing national systems such as Germany's.

So the honest summary is this: the pictogram-and-QR-code system that features in most illustrations of the EU's packaging future does not yet legally exist in usable form, while the regulation containing it applies from tomorrow. The delegated-act machinery is beginning to turn — a decision exempting certain pallet wrappings and straps from a reusable-transport-packaging requirement has been adopted — but the flagship provisions are behind.

2. Item-level identification, and what it would and would not enable

The technology most often described as transformative for EPR is digital watermarking: imperceptible codes covering a package's surface, readable by detection units on a sorting line, carrying data about material composition, food-contact status and producer identity.

The serious evidence comes from HolyGrail 2.0, an initiative of AIM – European Brands Association with the Alliance to End Plastic Waste, using Digimarc watermarking and Pellenc ST detection hardware. Its industrial trial ran for a hundred days from August to December 2024 at a materials recovery facility in Swisttal, Germany, processing material from around three million households. Reported results: 5.66 million detections across roughly 5,949 unique product SKUs, averaging about 56,000 detections a day, with single-pass detection efficiency of 87.9% to 93.8%. Brand participants included Aldi, Arla Foods, Mondelēz, Procter & Gamble, Coca-Cola and McDonald's. A France-wide pilot on rigid HDPE involving Henkel, L'Oréal, P&G, Veolia and Citeo has been announced, and the initiative is reported to be moving into a successor phase aimed at demonstrating economic viability.

Those are real figures from a real facility, and they are considerably better than sceptics predicted. Two qualifications are nonetheless essential. They are industry self-reported — the trial's results come from the participants' own reporting, not from independent academic or regulatory audit. And no jurisdiction has mandated digital watermarking, or made EPR fee modulation contingent on it. PPWR's digital-marking language is technology-neutral and names no technology; the choice is left to implementing acts that are themselves delayed. The United Kingdom's recyclability assessment methodology, which does drive modulated fees from October 2026, makes no mention of watermarks or QR codes.

The larger question is what item-level identification would mean for EPR's structure. The intuition is powerful: if every package can be identified at the sorting line, a producer could in principle be charged for its own actual items recovered, rather than for a share of a collective pot. That would be genuine individual producer responsibility, the arrangement Theme 1 identifies as theoretically preserving the design signal that collective schemes dilute.

Here the honest report is that the analysis does not exist. No academic work modelling item-level producer billing for packaging could be located. The closest relevant paper — Ellsworth-Krebs and colleagues (2022) on track-and-trace for reusable packaging — argues that current schemes penalise reuse because producers are charged by weight regardless of whether an item is reused, and proposes digital tracking to let governments exempt reusable packaging from per-fill fees. That is reuse-contingent fee modulation, not individual billing. And the same paper is explicitly sceptical of a fully individualised model, endorsing the view that circularity requires "not that one company closes the loop, but the ecosystem does."

The claim that digital identification will enable individual producer responsibility is therefore a vendor and consultancy proposition, not a research finding, and should not be attributed to any named scholar. It may well be right. Nobody has done the work.

3. Automated sorting: demonstrated against claimed

Sorting technology is where the gap between measured and marketed performance is widest, and the discipline of separating them matters because recycling rates are ultimately determined at the facility, not in the statute.

Peer-reviewed, real-facility evidence is modest in scale. A 2026 study in the International Journal of Advanced Computer Science and Applications evaluated an AI-driven robotic system in materials recovery facilities in Jakarta and South Tangerang, reporting 90% average sorting accuracy against 75% for manual sorting, 95.1% material purity against 84.9%, and throughput of 50 items per minute against 45. Those figures come from a controlled validation batch of 300 items — 100 PET bottles, 100 glass bottles, 100 aluminium cans — and the authors explicitly caveat that "actual MSW streams are more heterogeneous and unpredictable." The reported payback period is specific to Indonesian wage conditions and does not generalise.

Vendor claims are an order of magnitude larger and unaudited. AMP Robotics, via a case study hosted by the US Plastics Pact, claims throughput upwards of 80 items per minute, classification of more than 100 categories, and processing of more than 10 billion items annually across roughly 200 deployments. No independent testing body, audit date or facility-level verification is cited for any of it.

Independent survey work is more sobering. Lubongo and Alexandridis (2022) concluded that commercial sorting equipment addresses some materials recovery facility challenges but that film, multilayer, blended and mixed-material plastics remain problematic, because most equipment is designed for single-component streams. Laboratory work reporting perfect resin identification using spectroscopy and machine learning is laboratory work, not operational performance.

No large-scale independent assessment of AI sorting accuracy at commercial-scale American or European facilities was located. This matters directly for EPR because the difficult fraction — flexibles, laminates, multilayer — is precisely the fraction driving the plateau in reported rates, and it is precisely the fraction the survey literature identifies as unsolved.

4. Chemical recycling: contested, and being decided now

Whether chemically recycled material counts toward recycling and recycled-content obligations is the single most consequential unresolved definitional question in packaging policy, because it determines whether the difficult fraction can be counted as solved.

In the European Union, the position moved in mid-2026. Commission Implementing Decision (EU) 2026/1425 of 30 June 2026 permits chemical recycling to count toward recycled-content obligations under the Single-Use Plastics Directive, defining recycled plastic to include material from processes "during which the chemical structure of the material is changed," and establishing a mass-balance accounting methodology for cases where recycled input cannot be physically traced — with steam-cracker allocation rules, maximum three-month accounting periods, and no negative balances permitted. A later effective date applies to recycling in third countries.

Two cautions. This governs recycled content in single-use plastic beverage bottles under a specific directive; whether PPWR's own recycling-rate calculation rules treat chemical recycling the same way was not established and should not be assumed. And NGO commentary characterising the approach as embracing technologies with "harmful environmental impacts and low yields" is contested advocacy, not settled technical assessment.

In the United States the question is in active litigation. California's SB 54 regulations do not contain an unambiguous statutory inclusion or exclusion. The dispute centres on the regulations' treatment of hazardous waste from recycling facilities — language stating that hazardous waste managed under existing permits "is not automatically considered a large environmental or public health risk," which environmental groups argue could allow pyrolysis to qualify contrary to legislative intent. Those groups sued CalRecycle in June 2026 (the litigation article).

For EPR specifically, the stakes are structural rather than technical. If chemical recycling counts, the difficult fraction becomes a target-compliance problem solvable by capital investment. If it does not, the same fraction remains a design problem solvable only by not making the packaging — which is the source-reduction route (Theme 5).

5. Reuse: binding targets, aspirational targets, and one instructive failure

Reuse is where the binding/aspirational distinction does the most work, because the headline figures quoted in commentary mix the two indiscriminately.

Under PPWR, the 2030 figures are binding — 40% of transport and e-commerce packaging in reusable formats, and at least 10% of alcoholic and non-alcoholic beverages made available in reusable packaging, binding final distributors, with exemptions for highly perishable beverages, milk products, wine and spirits. The 2040 figures are not. Transport packaging at 70% and beverages at 40% are expressed as endeavour targets: the operative language for beverages is that distributors must "endeavour to make at least 40%" reusable, against "shall ensure" for 2030. A separate obligation requiring hospitality operators to let consumers bring their own containers at no extra charge is reported as binding from February 2027.

The instructive case is France, which offers something the EU's targets do not yet have: a track record. The AGEC law sets a national target that 10% of packaging placed on the market be reusable by 2027. According to ADEME's national reuse observatory, reusable packaging represented 1.82% of packaging placed on the market in 2024 — 2.75 billion units. France is not marginally behind its target; it is roughly a fifth of the way there with three years to run, having also implemented binding sector bans on disposable containers in food delivery, collective catering and school catering.

That gap between a legislated reuse target and observed uptake is the most useful available predictor of how PPWR's own reuse provisions will land, and it is more informative than any modelling.

In the United States, no binding percentage-based reuse mandate for packaging was located. What exists is permissive: Illinois, Oregon and Maine have removed health-code and liability barriers to consumer-owned containers. These are enabling statutes, not mandates, and the distinction should not be blurred.

6. Recycled content, and where it collides with EPR

Recycled-content mandates are converging with EPR from a different direction, and the interaction is under-examined.

PPWR sets minimum post-consumer recycled content in plastic packaging from 1 January 2030 and again from 2040, with separate categories for contact-sensitive PET, other contact-sensitive plastic, single-use beverage bottles and other plastic packaging. The precise percentages should be checked against the consolidated Annex before being relied upon: sources consulted for this article disagreed on the 2040 figure for contact-sensitive PET (50% against 65%) and on the 2030 beverage-bottle figure (30% against 35%). Only post-consumer recyclate counts. A verification and calculation methodology implementing act is required, and is itself pending.

In the United States, mandates run through state law rather than EPR statutes. California requires plastic beverage containers under its redemption programme to meet 15% recycled content from 2022, 25% from 2025 and 50% from 2030. Washington phases beverage containers to 50% by 2031, with separate schedules for household and personal-care containers and trash bags. Connecticut requires 25% by 2027 rising to 30% by 2032. New Jersey has requirements whose interim percentages could not be reconciled across sources and are therefore not stated here.

The interaction question is the interesting one. A producer facing both an EPR fee modulated on recyclability and a statutory recycled-content floor is subject to two instruments aimed at overlapping behaviour, with no coordination between them (the mandates article). No American state was found to have linked its EPR fees to recycled-content compliance in the way the United Kingdom links fees to recyclability grading. That is a design space nobody has entered.

7. The global treaty, which has not happened

Commentary frequently anticipates a global instrument containing EPR obligations. As of 11 August 2026, there is no adopted global plastics treaty, and the process has stalled repeatedly.

The Busan session in late 2024 ended without agreement. INC-5.2 in Geneva, 5–14 August 2025, ended without agreement, with unresolved splits over binding versus voluntary obligations, whether to address production as well as waste, the scope of bans and phase-outs, and finance. INC-5.3, in Geneva on 7 February 2026, was procedural only — convened to elect a new chair, with Ambassador Julio Cordano of Chile chosen. No text was negotiated. A further substantive session was expected later in 2026, but no confirmed date or location was found for this article.

EPR is an active subject of advocacy within the process: Citeo's director general has argued for a dedicated treaty annex on EPR with minimum requirements and national monitoring; South Africa's lead negotiator has emphasised integrating informal waste-picker labour and setting minimum standards to prevent regulatory arbitrage; the Business Coalition for a Global Plastics Treaty has called for national EPR systems built on common definitions. But this is advocacy about what should be in a treaty, not a description of drafted text. No binding EPR obligation currently exists in any agreed instrument, and none should be implied.

8. Where the argument stands

Four judgements follow from the survey.

The most consequential near-term change is definitional, not technological. PPWR's harmonisation of producer definitions, registration formats, recyclability grading and modulation bases across twenty-seven member states will do more to change how EPR operates than any sorting technology, because it determines what the numbers mean. That it is arriving with its implementing acts unwritten is the story of the next two years.

Item-level identification is the genuine structural possibility, and it is furthest from realisation. Digital watermarking works better than expected in an industrial setting on self-reported figures. Nobody has mandated it, nobody has modelled what it would do to EPR's collective architecture, and the one relevant academic paper argues against the individualised model it is assumed to enable. This is the field's largest gap between confident assertion and available evidence.

The chemical recycling question will be decided by regulators and courts within two years, and it determines everything downstream. Whether the difficult fraction can be counted as recycled is the difference between EPR as a capital-investment problem and EPR as a design problem.

Reuse is where legislated ambition and observed behaviour diverge most. France's 1.82% against a 10% target is the most instructive datum in this article, because it is the only case where a binding reuse target has run long enough to be assessed.

The connecting thread is that the next decade of EPR will be shaped less by what becomes technically possible than by what becomes legally countable. Every item in this article — watermarks, mass balance, endeavour targets, recycled-content verification methodologies — is ultimately a question about what the accounting will recognise. That is the same problem Theme 6 identifies at the heart of the field's measurement difficulties, arriving in new equipment.

References

  • EU instruments: Regulation (EU) 2025/40 (PPWR) — entry into force 11 February 2025, general application 12 August 2026; labelling and digital-marking provisions, reuse targets, recycled-content requirements and recyclability grading, with article numbering and exact percentages drawn from secondary analysis rather than a directly-read consolidated text, and flagged in the article where sources disagreed; Commission confirmation (June 2026) that the labelling implementing act deadline would be missed; Delegated Decision (EU) 2026/429 on pallet wrapping and strapping exemptions. Regulation (EU) 2024/1781 (ESPR) and the first working plan 2025–2030, which excludes packaging. Commission Implementing Decision (EU) 2026/1425 of 30 June 2026 on chemical recycling and mass balance under Directive (EU) 2019/904.
  • Digital watermarking: HolyGrail 2.0 industrial trial, Hündgen Entsorgung facility, Swisttal, 19 August – 19 December 2024, as reported by AIM – European Brands Association and the Alliance to End Plastic Waste and covered by Packaging Europe — industry self-reported, not independently audited; the France-wide HDPE pilot announcement. GS1's Sunrise 2027 is an industry initiative using "should" rather than "shall," not a mandate.
  • Individual producer responsibility: Ellsworth-Krebs, K. et al. (2022), "Circular economy infrastructure: Why we need track and trace for reusable packaging," Sustainable Production and Consumption 29: 249–258, including its endorsement of ecosystem rather than individual closure. No academic work modelling item-level producer billing was located.
  • Sorting: Nurhaida, I., Nasucha, M. and Nugraha, H. (2026), "AI-Driven Robotic Waste Sorting," International Journal of Advanced Computer Science and Applications 17(2) — controlled 300-item validation batch, with the authors' own heterogeneity caveat; AMP Robotics figures via a US Plastics Pact case study — vendor-reported, unaudited; Lubongo and Alexandridis (2022) on the limits of commercial equipment for film and multilayer plastics.
  • Reuse: PPWR reuse targets and the binding "shall ensure" (2030) versus "endeavour" (2040) distinction, cross-validated across two independent analyses; France's loi AGEC 10%-by-2027 target and ADEME National Reuse Observatory figure of 1.82% for 2024; US permissive statutes in Illinois (2023), Oregon (2023) and Maine (2024).
  • Recycled content: CalRecycle minimum content standards for redemption-programme beverage containers; Washington and Connecticut schedules. New Jersey's interim percentages could not be reconciled across sources and are not stated.
  • Global treaty: INC-5.1 (Busan, 2024); INC-5.2 (Geneva, 5–14 August 2025); INC-5.3 (Geneva, 7 February 2026, procedural, electing Ambassador Julio Cordano of Chile). Expert advocacy on EPR's inclusion via World Economic Forum coverage (November 2024), including Jean Hornain (Citeo), Mamogala Musekene (South Africa) and the Business Coalition for a Global Plastics Treaty.

Verification note: this article separates what is in force, what is legislated with a date, what is proposed, and what is vendor claim, because that distinction is routinely lost in commentary on this subject. Where EU article numbers and percentages rest on secondary paraphrase rather than a directly-read consolidated legal text, that is stated; the PPWR recycled-content percentages in particular carry documented source disagreement and should be checked against the consolidated Annex before being relied upon. Where no confirmed date exists — the next substantive treaty session — none is given. See Sources and method.