Grenfell Tower made combustible cladding a building-safety accountability test

Summary

• A fire that began in the kitchen of Flat 16 at 00:54 on 14 June 2017 escaped into the refurbished external wall. Aluminium composite material panels with a polyethylene core supplied most of the available fuel and were the principal reason for rapid spread. Combustible insulation, window-reveal materials, cavity geometry and defective or missing cavity barriers contributed to a system that did not adequately resist fire spread.
• The Grenfell Tower Inquiry found that the disaster was the product of failures over decades. Central government did not resolve known weaknesses in regulation and guidance; manufacturers distorted product evidence; testing and certification bodies failed to exercise adequate independence; designers, contractors, the cladding specialist and building control did not establish that the wall complied; the landlord and managing body did not manage fire safety or resident concerns adequately; and the London Fire Brigade was not prepared to reverse a stay-put strategy or handle the resulting volume of calls.
• Practical control was distributed but identifiable. Arconic controlled claims and disclosures about Reynobond 55 PE. Celotex and Kingspan controlled claims about their insulation. Studio E controlled external-wall design; Rydon controlled principal-contract delivery; Harley controlled specialist facade work; Exova controlled its fire-engineering advice; RBKC building control controlled regulatory scrutiny; the TMO and RBKC controlled landlord governance and resident information; and government controlled the regulatory framework. Residents controlled none of those functions.
• The Inquiry's findings are authoritative findings of a statutory public inquiry, not criminal convictions or civil liability judgments. As of 15 July 2026, the Metropolitan Police was still preparing and submitting files to the Crown Prosecution Service, and the CPS had not announced charging decisions. Corporate denials and remediation statements are relevant primary disclosures of company positions, but they do not displace the Inquiry's contrary findings.
• Repair evidence is substantial but incomplete. Parliament enacted the Fire Safety Act 2021 and Building Safety Act 2022; a Building Safety Regulator, accountable-person regime, safety case reports, stronger fire-service information duties and developer remediation obligations now exist. By 31 May 2026, however, government data showed that only 1,672 of 4,411 monitored residential buildings over 11 metres with unsafe cladding were recorded as complete, and that category included buildings awaiting building-control sign-off.
• A durable accountability test requires evidence at building level: a complete record of products and changes, competent design and review, tested system performance, installation inspection, named dutyholders, accessible resident concerns, evacuation planning, fire-service information, independent completion evidence and disclosure of remaining risk. A statute, funding commitment or reported programme milestone is not by itself proof that a particular occupied building is safe.

The confirmed event and the boundaries of the record

The starting point is the two-stage statutory inquiry, not later summaries by organisations whose conduct it examined. The Inquiry's Phase 1 report reconstructed the night, including the origin of the fire, escape into the external wall, spread around the tower and emergency response. Its Phase 2 overview addressed the longer path to disaster: regulation, product testing and marketing, certification, the refurbishment, management of the tower, the fire service, response and recovery, and reform. Phase 2 also revisited technical questions left provisional in Phase 1.

Several facts are firmly established. The fire started at about 00:54 in the kitchen of Flat 16 on the fourth floor. It escaped through the window area and reached combustible materials in the external-wall system. It then travelled rapidly up, down and around the 24-storey residential tower, re-entering flats and destroying the compartmentation assumptions on which the stay-put strategy depended. Seventy-two people died. The last survivor escaped at 08:07. Those facts are not in serious evidential doubt.

The technical record also supports a hierarchy of contribution. The rainscreen panels were Reynobond 55 PE, an aluminium composite material made from a three-millimetre polyethylene core between thin aluminium sheets. The Inquiry found that these panels acted as fuel and were the principal reason for the speed and extent of external spread. Its later experimental work, described in Volume 7, confirmed that ACM PE had by far the largest potential contribution to energy release in the wall. The experiments also showed why it is incomplete to describe the case as a panel-only failure: insulation retained heat and could burn, panel deformation exposed the core, molten polyethylene moved and pooled, and the cavity geometry affected heat loss and flame development.

The external wall contained mainly Celotex RS5000 polyisocyanurate insulation and a smaller amount of Kingspan K15 phenolic insulation. Both were combustible. The Inquiry did not find that changing only the insulation would have removed the extreme hazard created by ACM PE. It did find that insulation was needed to retain energy or add energy for full system involvement, and that the wall as a whole failed to resist fire spread. That distinction matters because product accountability and fire causation are related but not identical.

A manufacturer's dishonest marketing can be a grave institutional failure even if another product contributed more energy to this particular fire.

There were additional routes and defects. The refurbishment moved windows outward and introduced combustible uPVC and insulation in the reveals. Phase 2 confirmed the earlier conclusion that deformation and loss of the window surround created a route from Flat 16 into the cladding cavity. The design and installation of cavity barriers were deficient; barriers were absent around window openings, some were incorrectly positioned, and products were used in voids outside tested conditions. Fire could bypass intended divisions. These conditions were not a random coincidence on the night.

They were outcomes of design, specification, substitution, coordination, construction and inspection decisions.

Other facts require careful limits. The Inquiry made findings about conduct and institutional responsibility under its terms of reference. It did not determine criminal guilt, sentence individuals, adjudicate damages or allocate contribution percentages among defendants. Its conclusions can inform later legal processes, but those processes have their own standards and admissible evidence. Likewise, a civil settlement is evidence that claims were resolved, not an admission or judicial finding unless its terms say so.

Keeping these boundaries visible is essential in a case where public findings, criminal investigation, confidential settlements and corporate denials coexist.

Practical control was fragmented, but not unknowable

The phrase "everyone was responsible" can become a way of making no one answerable. Grenfell requires a more precise control map. Central government controlled the Building Regulations, Approved Document B and the policy environment in which construction products and building control operated. It could change statutory guidance, clarify the treatment of combustible materials, respond to coroner and select committee recommendations, collect information about facade fires and regulate construction products. It did not select Grenfell's panels, but it controlled the rules against which that selection was judged.

Product manufacturers controlled a different evidential layer. Arconic Architectural Products SAS manufactured and supplied the Reynobond 55 PE sheets used to fabricate the rainscreen. It controlled the testing information, classifications, warnings, marketing and communications attached to that product. Celotex controlled the development and promotion of RS5000. Kingspan controlled claims and certificates used to market K15. None of those companies designed the entire Grenfell wall. Each nevertheless controlled whether customers and certification bodies received accurate, complete and appropriately bounded evidence about its own material.

Testing, accreditation and certification bodies occupied the assurance layer. BRE conducted system tests and produced reports. The British Board of Agrement issued product certificates. Local Authority Building Control produced certificates used in the market. UKAS accredited certification and testing activity. Assurance was supposed to challenge supplier claims and preserve the scope of test results. Where the customer controlled the specimen or a report omitted a safety-critical modification, the value of the assurance depended on the body detecting and disclosing that fact.

The refurbishment team controlled translation from products into a building. Studio E, as architect, was responsible for the external-wall design and material selection. Rydon, appointed as design-and-build principal contractor, controlled delivery and coordination after appointment. Harley Facades was the specialist cladding subcontractor and controlled important detailed design, procurement and installation work. Exova provided fire-engineering services, although its retainer became unclear and its fire strategy was not completed. Artelia performed client-side project functions.

RBKC building control had statutory authority to examine compliance and ultimately accepted the work.

The client and landlord side also had real powers. The Royal Borough of Kensington and Chelsea owned the tower. The Kensington and Chelsea Tenant Management Organisation managed it and acted as refurbishment client. The TMO controlled professional appointments, cost decisions, resident communication, fire-risk management, maintenance, emergency planning and the escalation of complaints. The Inquiry's criticism was not that housing managers should have recalculated every facade test.

It was that the client failed to secure competent advice, clarify responsibility, complete the fire strategy and give sufficient weight to safety and residents.

The London Fire Brigade controlled operational preparation and response. Before the fire it could inspect, issue deficiency or enforcement notices, gather risk information and prepare high-rise procedures. During the fire it controlled incident command, operational deployment, communications and the advice passed through control-room calls. It did not create the combustible wall. Once compartmentation failed, however, its ability to recognise that stay put was no longer tenable and to coordinate evacuation became safety-critical.

Residents had the smallest share of institutional control and the largest direct exposure. They could complain, attend meetings, report defects and decide whether to leave when conditions allowed. They could not inspect concealed materials, demand a large-scale system test, reject a substitution, complete the fire strategy, overrule building control or redesign the single stair. Treating residents as another equal actor in a fragmented chain would erase the asymmetry that made their warnings important: they were supplying local risk information to organisations that held the authority and technical resources to act.

The path to disaster began before the refurbishment

The Inquiry's Volume 1 account of the path to disaster traces relevant facade fires and warnings through Knowsley Heights in 1991, Garnock Court in 1999 and Lakanal House in 2009. The precise construction and fire dynamics differed, so none was a perfect preview of Grenfell. Together they showed recurring hazards: combustible external materials, hidden cavities, missing barriers, fire spread around windows, unclear guidance and the vulnerability of high-rise evacuation assumptions when compartmentation failed.

Government had opportunities to improve the framework. A parliamentary committee warned in 1999 about combustible materials in external walls. The Lakanal House coroner made recommendations after six deaths in 2009, including clearer guidance and better information for residents. The Inquiry found that important recommendations were not acted on with the necessary urgency. It described a departmental culture in which deregulation pressures, fragmented responsibility and weak escalation made decisive action unlikely without a major fatal fire.

Approved Document B contained two fundamental problems. First, its wording and structure did not communicate clearly how functional requirements should be applied to the external wall of a high-rise residential building. Industry entities often treated the guidance as if it were the regulation itself, yet also interpreted individual passages in isolation. Second, the Class 0 surface classification route did not measure the behaviour that mattered in a facade fire. A material could satisfy surface-spread tests that did not expose a combustible core or reproduce an external-wall cavity, deformation and full-system interaction.

Large-scale BS 8414 testing offered system evidence, but its results applied to the configuration tested. They were not a certificate that one product could be inserted into any wall. That limitation was repeatedly blurred in the market. A test of one build-up, with defined fixings, barriers, insulation, panel arrangement and fire-stopping, could not honestly establish that an insulation board or panel was universally suitable above 18 metres. Grenfell demonstrates why evidence portability is a control in its own right: whoever relies on a test must show that the proposed construction falls within its field of application.

The regulatory structure diluted ownership. Building regulations sat with one department, product regulation with another and fire services with another. Local authority building control competed with private approved inspectors in parts of the market. Certification bodies had commercial relationships with manufacturers whose products they assessed. The Hackitt independent review, published after the fire but before Phase 2, described a system not fit for purpose, with unclear roles, inadequate oversight, weak enforcement and a culture of doing enough to pass responsibility onward rather than demonstrating safety.

This does not mean every pre-2017 use of combustible material was knowingly unlawful or that every official understood the mechanism later reconstructed. It means warning evidence and known uncertainty were not converted into precaution, clear rules and competent verification. The supported inference is that an effective learning system would have removed or sharply constrained the interpretive routes used at Grenfell before the refurbishment. The counterfactual cannot identify the exact date or wording that would have prevented every unsafe choice.

Product evidence became a control failure of its own

The Inquiry's Volume 2 is the primary record for product testing, marketing and certification. Its findings were severe. It found systematic dishonesty by manufacturers, including manipulation of test processes and calculated efforts to lead the market to believe combustible products met guidance. It also found that certification and testing organisations did not apply the independence and rigor required to expose those claims.

For Reynobond 55 PE, the Inquiry found that Arconic possessed test results showing materially different fire performance depending on panel configuration and had information relevant to high-rise use. It found that the company continued to sell the PE product for architecture and did not give the BBA or the Grenfell project a candid account of the danger and test evidence. The panel's polyethylene core was not an incidental adhesive. It was a substantial combustible layer protected by thin aluminium skins that could deform and separate under fire.

Arconic disputes the Inquiry's characterization. Its September 2024 company statement says the product was legal to sell, safe as a building material, tested by third parties and not sold with concealed information. That statement is useful evidence of the company's position and acknowledgment that its subsidiary supplied the sheets. It is not independent technical validation. The Inquiry examined the underlying documents and testimony and reached contrary findings. An accountable analysis records both without treating them as equivalent in evidential weight.

For Celotex RS5000, the Inquiry found a dishonest scheme to enter the high-rise insulation market. A 2014 BS 8414 test build included magnesium oxide boards in critical locations, but the resulting report did not disclose them. Celotex then marketed RS5000 as the first PIR board successfully tested to BS 8414 and acceptable above 18 metres, even though the test applied to the system and had been manipulated. The omission severed the commercial claim from the specimen that passed.

For Kingspan K15, the Inquiry found a long-running false market based on claims that success in one BS 8414 system made K15 generally suitable for buildings over 18 metres. It found reliance on inaccurate certificates and tests of modified or trial products rather than the product then sold. Kingspan's Grenfell Inquiry response site accepts historical conduct and process shortcomings and describes board oversight, compliance certification and other changes. It argues that K15 was only a small part of Grenfell's insulation and did not materially alter spread driven by ACM PE. The first proposition about quantity and the Inquiry's technical ranking can be true while the product-marketing findings remain valid.

The British Board of Agrement, BRE and Local Authority Building Control were not mere conduits. Their certificates and reports enabled downstream reliance. The Inquiry found that the BBA did not examine product evidence with sufficient care, BRE permitted manipulation and issued an incomplete Celotex test report, and LABC certificates adopted manufacturer language without adequate technical scrutiny. UKAS did not identify important weaknesses in the accreditation process. These were assurance failures because the organisations existed to reduce the information advantage held by manufacturers.

The durable lesson is not that every material must be judged from a single test label. It is that claims need traceability from product identity through specimen, method, configuration, result, classification, permitted field of application and actual installation. A change to any safety-critical element must trigger review. Marketing language cannot enlarge a test result. Certification bodies must be able to withstand commercial pressure, and purchasers must obtain the full evidence rather than a logo or summary phrase.

The 2012-2016 refurbishment converted ambiguity into an unsafe wall

Grenfell Tower was a reinforced-concrete residential building completed in the 1970s. Its original concrete exterior was not combustible. The 2012-2016 refurbishment added a ventilated rainscreen system, insulation, replacement windows and related works. The project sought energy, weather and appearance improvements and was linked to wider estate development. Cost pressure influenced the selection of ACM panels. Cost control is ordinary in public construction; the accountability issue is whether savings were subjected to a non-negotiable safety gate.

The Inquiry's Volume 4 reconstruction of the refurbishment found that no entity demonstrated the compliance of the external wall as a complete system. Studio E specified combustible materials and failed to recognise the danger that a reasonably competent architect should have recognized. Rydon did not take effective responsibility for design after becoming design-and-build contractor. Harley, despite specialist facade expertise, did not establish that the materials and details were suitable. Exova's fire strategy remained incomplete and its continuing scope was never clarified. RBKC building control approved the work without resolving fundamental questions.

Responsibility was not absent from the contracts; it was poorly understood and repeatedly passed sideways. One entity assumed another had checked the insulation. Another treated building control comments as proof of design. Product literature stood in for system analysis. Detailed drawings evolved without a single controlled fire-safety review. The Inquiry found an unacceptably casual approach to contractual relations and no adequate understanding of the Building Regulations among key project staff.

The window reveals show how several small decisions formed a dangerous interface. Windows were moved outward into the line of the new cladding. Combustible uPVC boards and insulation were used around them. Proprietary cavity barriers were not installed around every opening. The fire in Flat 16 did not have to breach a robust non-combustible boundary; hot gases and flames encountered materials and spaces that provided access to the cavity. An interface owned by no one in practice became the route into the system.

Cavity barriers provide a similar control test. A barrier that has passed an isolated resistance test may not perform in a ventilated facade where panels deform, cavities exceed the tested width or fixings and rails create bypass routes. Siderise supplied barriers for use in some voids larger than tested, although the Inquiry found no dishonesty by the company. The design team and installer still had to match each product to the actual cavity, orientation and support. Building control had to verify the evidence. Those steps did not happen reliably.

RBKC building control's involvement did not transfer responsibility away from designers and contractors. Building control was a regulatory check, not the project's design office. Equally, the professionals' duty did not excuse weak scrutiny by building control. Multiple controls should have been independent barriers. Instead, they became circular reliance: the designer expected the specialist and fire engineer to identify a problem; the contractor expected the designer and building control; the specialist relied on product literature; and building control relied on submissions that never proved the whole wall.

The TMO shared responsibility as client. It had no experience of overcladding a high-rise tower and did not compensate for that limitation by establishing a strong, independent safety assurance process. It failed to ensure that Exova's position was clarified after Rydon's appointment and that a final fire strategy existed. It also pursued cost reductions without securing evidence that the resulting wall remained compliant. The failure was not choosing a lower price as such. It was allowing commercial selection to close before safety evidence did.

Resident warnings exposed governance weakness before the fire

The Inquiry's Volume 3 examined the TMO's relationship with residents and its management of fire safety. It found a relationship marked by mistrust, defensiveness and hostility. Residents raised concerns about refurbishment quality, smoke ventilation, fire instructions, access and other conditions. Some concerns did not identify the specific ACM PE mechanism later established. Their value was still substantial: they showed that people living with the works did not trust the managing organisation to investigate hazards or communicate candidly.

After a 2010 fire, smoke spread through lobbies and residents questioned the smoke ventilation system and the TMO's approach. In March 2014, the LFB issued a deficiency notice after finding that about a quarter of smoke-control dampers were not working. A replacement system formed part of the refurbishment, but the history showed that a central safety system needed close commissioning and maintenance evidence. Resident reports were not isolated from technical control; they were another signal about whether those systems worked in the occupied building.

Fire-safety information was inconsistent. A new resident handbook took years to finish. Advice appeared irregularly in general newsletters. There was no reliable system proving that each new resident received or understood the relevant instructions. In November 2016, the LFB issued a deficiency notice that included the absence of fire action notices. When resident Edward Daffarn publicly criticised the TMO's fire-safety management, an internal reaction included asking whether the criticism was libellous rather than first examining the underlying risk. The Inquiry treated that reaction as evidence of institutional defensiveness.

The TMO also had information about vulnerable residents, but the record used on the night was incomplete. A 2014 report identified 57 residents with sensory, cognitive or mobility impairments, yet there is no evidence that the information was used to update the operational spreadsheet later searched during the fire. A premises information box requested by the LFB was not available in the form needed. These gaps mattered because a stay-put building with one stair requires rapid knowledge of who cannot self-evacuate once the strategy changes.

It would be inaccurate to say residents predicted the precise fire or possessed proof that the external wall would fail. The Inquiry itself cautioned against turning a famous 2016 blog post into a technical prediction. The confirmed point is stronger and more useful: residents correctly identified that fire safety was not being managed with the seriousness, openness and responsiveness required. The governance system did not convert their lived evidence into an independent technical audit.

An accountable complaints process must do more than acknowledge correspondence. It should classify safety allegations, preserve them, link repeated signals across properties, assign a competent investigator, state the evidence reviewed, track corrective action and allow escalation outside the landlord. The later Building Safety Act's resident engagement and complaints duties address part of that gap. Their effectiveness must be measured by outcomes: response time, independent review, closed hazards and resident access to evidence.

A domestic fire became a building-wide emergency

At 00:54 on 14 June 2017, the first call reported a fire in Flat 16. Firefighters arrived quickly and controlled the compartment fire. By about 01:09, flames had escaped through the window area into the external wall. The external fire then accelerated beyond the assumptions under which the building and response plan operated. Within a short period it reached upper floors, travelled laterally and re-entered flats through windows.

Stay put was not inherently irrational for a concrete high-rise building with reliable compartmentation. The strategy seeks to keep residents away from smoke in common escape routes while firefighters contain a fire in the flat of origin. At Grenfell, the new wall allowed external spread and multiple re-entry, flat doors and other barriers did not all provide the expected protection, and smoke entered lobbies and the single stair. The premise supporting stay put was disappearing while advice continued to rely on it.

The Inquiry found that the London Fire Brigade was not prepared for this transition. Its high-rise policy did not include a workable plan for full or partial evacuation. Incident commanders had not been adequately trained to identify when stay put had failed or how to change strategy. Risk information about combustible facades held by specialists did not reach operational policy and training. Radio communication was unreliable in the reinforced-concrete tower. Command, deployment and information systems were overwhelmed.

Control-room staff faced an unprecedented volume of fire-survival-guidance calls. Training had not prepared them to manage many simultaneous callers trapped in different flats. Information passed from callers to the incident ground through improvised processes that could not reliably track changing conditions, rescue priorities and completed deployments. The general change from stay-put advice was not made until 02:35, long after external spread was evident. By then, conditions in the stair and many flats made evacuation more dangerous or impossible for some residents.

The Inquiry's Volume 5 examination of the LFB traced these weaknesses to management before the night, not simply decisions by individuals under extreme conditions. Lessons from the 2009 Lakanal House fire had not been embedded. Incident-command training and refresher assessment were deficient. Control-room functions were not integrated effectively. Building inspections did not create reliable operational risk records. Communications limitations were known but not adequately addressed.

Firefighters and control-room officers worked with courage in conditions no residential brigade should have faced. Institutional accountability does not require portraying frontline responders as indifferent. It requires distinguishing individual effort from the system that prepared, equipped and directed them. Heroic action can coexist with organizational failure, and emphasizing one must not erase the other.

The emergency response beyond firefighting also failed residents. More than 800 people were displaced from the tower and surrounding homes. RBKC's humanitarian response lacked effective leadership, information and coordination. Voluntary and community groups supplied food, shelter, local knowledge and practical assistance when official provision was inadequate. The Inquiry found that the community's response highlighted, rather than cured, the public system's deficiencies.

Impact cannot be reduced to one financial total

The primary impact was the loss of 72 lives and the injury, displacement and trauma experienced by survivors, bereaved families, residents, responders and the North Kensington community. The Inquiry devoted individual findings to the circumstances in which people died. That record rejects abstraction: building-safety controls were not paperwork failures detached from consequence. They determined whether people could remain protected in their homes, receive accurate advice, use a stair and be found by rescuers.

There were long-term health and social effects that public records cannot aggregate into a single verified measure. Survivors and families experienced bereavement, trauma, interrupted education, insecure housing and years of investigation and legal process. Responders were exposed to extreme events. Nearby residents lost homes or community stability. Public agencies incurred rehousing, site, inquiry, legal, health and remediation costs. Some harms are compensable; others cannot be restored by payment.

An alternative dispute-resolution process produced a civil settlement of around GBP150 million for survivors, bereaved families and residents. The department's 2022-23 accounts confirm the aggregate settlement and a government contribution while preserving confidential details. The amount is not a judicial valuation of every death or loss and does not establish each settling party's liability. Confidential settlement resolves claims without creating the same public factual record as a trial.

Corporate proceedings also continued. A May 2026 SEC-filed Howmet disclosure states that personal-injury claims by survivors and estates had settled and closed. It separately describes claims by RBKC and KCTMO against Arconic entities and Whirlpool, defenses and contribution claims, and a trial window then anticipated between October 2028 and July 2029. This is a company disclosure about procedural status, not a court finding on the merits.

The broader building-safety crisis created another category of impact. Leaseholders and tenants in other buildings faced unsafe homes, waking watches, insurance costs, delayed sales, mortgage constraints, service charges and prolonged uncertainty. Developers, landlords and taxpayers faced remediation obligations. Small contractors and property managers faced capacity and liability pressure, but smaller balance sheets do not remove life-safety duties. A fair system must distinguish responsibility from ability to pay while preventing costs from defaulting onto residents who did not create the defects.

The legal framework for recovery has changed. In URS Corporation v BDW Trading, the UK Supreme Court examined post-Grenfell remediation and the retrospective extension of Defective Premises Act limitation periods under the Building Safety Act. The judgment was not about liability for the Grenfell refurbishment. It confirmed that the Act was designed to improve routes to redress for historic building defects and addressed when a developer that undertook remediation could pursue its engineer. That distinction prevents a wider building-safety judgment from being misreported as a Grenfell merits ruling.

Public inquiry findings are not criminal verdicts

The Inquiry's final report named organizations and individuals, assessed documents and testimony and made strong findings. Those findings have substantial public authority. They still do not establish the elements of offences beyond reasonable doubt. Potential criminal cases may involve corporate manslaughter, gross negligence manslaughter, health and safety, fraud or other offences, but charging and trial require separate decisions about suspects, admissible evidence, public interest and legal tests.

As of the article's publication date, no criminal conviction had established responsibility for causing the Grenfell fire. The Metropolitan Police update of 19 May 2026 said 15 of 20 files had been submitted for early or charging consideration, 10 of 14 overarching evidential files were complete and all files were expected to be submitted by the end of September 2026. The investigation team had expanded to 220 people. The police expressly said that preparation for possible trials did not presume charges.

The Crown Prosecution Service statement likewise said specialist prosecutors would review the evidence independently and did not expect charging decisions before the end of 2026. That timetable is difficult for families who have waited since 2017, but it defines the confirmed legal position on 15 July 2026. It would be false to call organizations "convicted" on the basis of the Inquiry or to predict charges as certain.

There is also a difference between causation and culpability. The Inquiry could find that a product was the principal reason for fire spread and that a company concealed relevant information. A criminal court would still have to examine the precise defendant, offence, state of mind, duty, causation and available defenses. A professional failure can be a cause without satisfying a criminal threshold. Conversely, conduct that did not add much energy to this particular fire might still be legally significant if false statements or safety duties are proved.

Corporate statements need the same boundaries. Arconic and Kingspan have rejected or qualified important Inquiry conclusions while describing settlements and reforms. Their statements are relevant because accountability includes a right to respond and because they create commitments that can later be tested. They are not substitutes for external assurance. A company can acknowledge process shortcomings without accepting causation; an inquiry can find dishonesty without a criminal court having convicted the company. Both propositions can remain in the record.

Reform changed the architecture of responsibility

The first major post-fire review was Hackitt's. It proposed clear dutyholders, stronger competence, a more effective regulator, a golden thread of building information and resident participation. Parliament later enacted the Fire Safety Act 2021, clarifying that the Fire Safety Order applies to the structure, external walls, cladding, balconies and flat entrance doors of multi-occupied residential buildings. This closed an important ambiguity about who had to assess external-wall risk in occupation.

The Building Safety Act 2022 created a wider framework. It established the Building Safety Regulator, dutyholder and competence requirements for design and construction, an accountable-person regime for occupied higher-risk buildings, mandatory occurrence reporting, resident engagement duties, stronger redress and leaseholder protections, and powers concerning construction products. It also extended limitation periods and created mechanisms such as remediation contribution and building liability orders.

The occupied-building regime assigns a principal accountable person and other accountable persons according to legal control of common parts and structural obligations. For a high-rise residential building, the principal accountable person must register the building, coordinate risk management, maintain a resident engagement strategy, operate complaints and occurrence-reporting systems and prepare a safety case report. The regulator's safety case guidance requires construction and material information, structural condition, refurbishment and remediation history, fire measures, risk assessments, emergency planning and safety-management arrangements.

This model addresses a central Grenfell failure: evidence scattered across owner, manager, architect, contractor, product supplier and fire service. The golden thread is intended to keep accurate, accessible and controlled information through the building lifecycle. A safety case is not merely a document. It should be the reasoned demonstration that fire-spread and structural risks are identified and controlled. If it becomes another certificate purchased from a consultant and stored without operational use, the reform will reproduce the behavior it was intended to replace.

The Fire Safety (England) Regulations 2022 introduced duties for high-rise residential buildings to give fire and rescue services electronic plans and external-wall information, maintain secure information boxes, check firefighting lifts and equipment, install wayfinding signage and provide residents with fire-safety instructions. Buildings over 11 metres have recurring fire-door check duties. These controls map directly to information and equipment failures examined by the Inquiry.

Evacuation reform took longer. The Fire Safety (Residential Evacuation Plans) (England) Regulations 2025 came into force on 6 April 2026. Government evacuation guidance now addresses building-level plans and person-centred arrangements for residents who may have difficulty self-evacuating in specified buildings. The May 2026 government record still classified one element of the Inquiry's personal emergency evacuation recommendation as in progress pending primary legislation. "Implemented" must therefore be attached to the particular duty, not used as a blanket claim.

The London Fire Brigade says it completed all 29 Phase 1 recommendations directed to it by March 2024. Its published completion account identifies new high-reach appliances, command units, fire-survival-guidance arrangements, incident-command changes and training. Those are concrete changes. The evidence is primarily the Brigade's own report; durable assurance also requires inspection findings, exercise results, incident learning and proof that competence survives turnover and unusual demand.

Remediation evidence shows progress and an unfinished crisis

The physical tower is not a repaired residential building. It has not been reoccupied. Government took ownership of the site and decided in February 2025 that the structure would be carefully taken down. The official site decision describes progressive deconstruction and continuing engagement with the community and Memorial Commission. Site work is therefore preservation, deconstruction and memorial stewardship, not evidence that Grenfell itself was remediated for safe occupation.

The national repair question concerns thousands of other buildings. Government programmes address ACM cladding, other high-rise external-wall defects, mid-rise buildings, developer-led remediation and social housing. A developer remediation contract requires participating developers to assess and fix life-critical fire-safety defects in residential buildings over 11 metres that they developed or refurbished within the defined 30-year period. The Responsible Actors Scheme can impose development and building-control consequences on eligible developers that do not participate or comply.

The latest release available by 15 July 2026 was the May 2026 building-safety remediation data. It reported 4,411 monitored residential buildings over 11 metres with unsafe cladding, estimated to represent 61% to 76% of the buildings expected to enter the programmes. Of those monitored buildings, 1,672, or 38%, were recorded as complete; 659, or 15%, were underway; and 2,080, or 47%, had not started. The completion category included buildings waiting for building-control sign-off.

The ACM subset showed stronger progress. Of 516 identified high-rise residential and publicly owned buildings with ACM systems unlikely to meet regulations, 500 had started or completed remediation and 472 were reported complete. Fourteen of the 472 were waiting for building-control sign-off. Sixteen had not started, including occupied buildings with forecast dates or enforcement action. The numbers prove a major reduction in exposure to the panel type most closely associated with Grenfell. They also prove that the programme was not finished nine years after the fire.

Broader programme data is more difficult. Eligibility continues to identify additional buildings. A building may appear in more than one route. Developer reporting changed in April 2026 to separate cladding and non-cladding work, causing status reclassification. Some figures are self-reported. The government explicitly labels the release management information rather than official statistics. A reported decline in "started or complete" from one month can reflect better classification rather than work reversing. Honest accountability requires preserving those qualifications.

The May 2026 Inquiry progress report recorded 21 of 61 tracked recommendations as complete and 40 in progress. The count is 61 because it includes 58 Phase 2 recommendations plus three remaining Phase 1 matters concerning vulnerable people and evacuation. Government expected all recommendations to be implemented by the end of 2029, with some requiring legislation. It reported movement toward a single construction regulator, consultation on Approved Document B, a construction-products white paper, fire-risk-assessor certification proposals and work on the fire-engineering profession.

These are design and implementation signals, not equivalent outcomes. Publishing a consultation does not complete reform. Establishing a regulator does not prove it has the staffing, data, independence and enforcement capacity to examine every building. Signing a developer contract does not prove each defect is identified or each repair is complete. "Complete awaiting sign-off" is not the same as independently accepted completion. The public record needs a hierarchy: proposal, law, operational system, building-level action, independent verification and sustained performance.

What repair evidence should exist at building level

The first requirement is an accurate building record. It should identify every external-wall product, manufacturer, batch where available, fixing, cavity barrier, insulation type, window interface, fire stop and later change. Drawings must reflect the installed building, not only the tender design. Product certificates should link to full reports and valid fields of application. Where records are missing, intrusive inspection and sampling should replace assumption.

The second is a competent system assessment. The assessor should examine the wall as built, not classify products in isolation. The assessment should state fire scenarios, uncertainties, deviations from tested systems, compartmentation, evacuation assumptions and interim measures. Independence and conflicts must be disclosed. Peer review should be proportionate to consequence, particularly where an assessment allows combustible material to remain.

The third is controlled remediation design. Removal of ACM can uncover missing barriers, defective fire stopping, wet insulation, unsuitable membranes or structural issues. Scope must address life-critical non-cladding defects as well as the visible panel. Changes should be reviewed through the dutyholder process, and residents should receive a clear explanation of what is and is not being fixed. The government's developer guidance rightly states that an EWS1 form is a valuation tool, not a statutory safety certificate.

The fourth is construction evidence. Inspectors should record substrate condition, barrier continuity, fixings, cavity dimensions, window details and concealed work before closure. Photographs need location and date metadata. Nonconformities, approvals and rework should be traceable. Sample inspection without a risk basis can miss repeated concealed defects. Completion information must enter the golden thread and be given to the accountable and responsible persons.

The fifth is independent acceptance. Building-control sign-off is one element, not the whole safety case. Fire-risk assessment, external-wall appraisal, commissioning of alarms or smoke control, updated evacuation plans and fire-service information also matter. Where a programme reports completion before sign-off, it should publish both states. Where a developer's qualifying assessment says risk is tolerable, government audit and regulator review should test the conclusion.

The sixth is operational readiness. Accountable persons should update their safety case, resident engagement strategy, occurrence-reporting process and emergency plan after work. Fire and rescue services should receive accurate wall, floor and equipment information. Residents who may need assistance should have current arrangements with consent and privacy protections. A repaired wall does not cure a broken smoke system, fire door, single-stair management process or call-handling plan.

The seventh is continued assurance. Facades weather, seals fail, fixings corrode, residents alter doors and management companies change. Periodic inspection must be based on degradation mechanisms and safety significance. Complaints and near misses must feed back into risk review. The HSE-sponsored high-rise incident research supports scenario-based safety narratives, near-miss learning and strong dutyholder leadership. It is research input rather than proof that the statutory regime now performs those functions everywhere.

Confirmed repairs, supported inferences and missing evidence

Confirmed repair facts include enacted statutes, an operating Building Safety Regulator, registered higher-risk buildings, legal duties for accountable persons, fire-service information requirements, a developer contract and multiple remediation funds. Government publishes monthly programme data and quarterly Inquiry recommendation updates. LFB has changed policies, equipment and training and reports completion of Phase 1 actions. ACM removal has occurred on most identified high-rise buildings in the dedicated programme.

It is a supported inference that these changes reduce risk. A named principal accountable person should reduce ambiguity over occupied-building management. A safety case should reveal evidence gaps that a simple compliance certificate missed. External-wall information should help firefighters prepare. A ban and remediation of ACM PE should remove the dominant fuel mechanism seen at Grenfell. Mandatory reporting and resident complaints should create escalation routes. These are reasonable causal expectations grounded in the design of the controls.

The public evidence does not prove uniform effectiveness. There is no complete public register of every building over 11 metres with an unsafe wall and its verified status. Government estimates that the monitored 4,411 buildings cover only 61% to 76% of the expected programme population. Completion data combines signed-off and awaiting-sign-off buildings. Some status data is supplied by developers or housing providers. The public cannot inspect every safety case, field report, product record, enforcement file or resident complaint outcome.

Regulatory capacity is another unknown. The number of registrations does not show how deeply each safety case has been assessed. Published programme counts do not show whether every fire-service plan is current. Consultation on a single regulator, product reform, fire engineering and assessor competence shows that important parts of the architecture were still evolving in 2026. A regulator can exist in law before its systems, staffing and case throughput reach the intended level.

Corporate remediation evidence also remains bounded. Kingspan describes governance, compliance and testing changes; Arconic describes settlements and continuing legal engagement. Without public independent audit results, these disclosures establish commitments and company positions, not full assurance that incentives, product claims and escalation behavior have changed across all operations. The Inquiry's core lesson is precisely that polished certificates and marketing statements can conceal weak evidence.

Criminal accountability remains unresolved. Police and CPS work may lead to charges, no charges, or charges against some but not all investigated subjects. The public record cannot predict the outcome. Civil settlements resolve claims but confidential terms limit public learning. Continuing contribution litigation may allocate financial responsibility for some costs, yet a later judgment cannot restore lives or replace preventive control.

Finally, there is no valid counterfactual count of how many people would have survived under one isolated change. Non-combustible panels would have transformed external fire spread, but exact life outcomes would still depend on window details, insulation, barriers, doors, smoke, advice, stair conditions and rescue. Earlier evacuation might have saved more people, but ordering it at a particular minute could also have exposed residents to smoke. Counterfactuals must remain conditional.

Counterfactuals that identify controllable intervention points

The strongest counterfactual begins before procurement. If Approved Document B had clearly prohibited ACM PE and combustible insulation on high-rise residential external walls, the Grenfell design would have had no interpretive path to those products. This is strongly supported because non-combustible materials would have removed the principal facade fuel. It does not prove that every other defect would have been fixed or that no external fire could occur.

A second intervention is product evidence. If Arconic had supplied complete configuration-specific fire data and explicit high-rise warnings, if Celotex's magnesium oxide boards had appeared in the report, and if Kingspan claims had remained within tested systems, designers and certifiers would have faced a very different evidence set. Honest evidence does not guarantee a competent buyer, but it removes false assurance. Certification bodies applying rigorous challenge could have stopped the claims even if manufacturers did not.

A third intervention is a formal external-wall safety case at design freeze. One accountable lead could have been required to identify the proposed materials, demonstrate B4 compliance for the whole build-up, reconcile BS 8414 evidence, resolve window and cavity details, obtain independent fire-engineering review and record sign-off before purchase. Any substitution would reopen the case. Grenfell had many documents but no such integrated proof.

A fourth intervention is competent building control. An inspector could have demanded a completed fire strategy, valid system evidence and coordinated as-built drawings before accepting the wall. Rejection at any point would have delayed the project and increased cost. Those are appropriate consequences when evidence for a high-consequence facade is absent. The uncertainty is whether the same market misunderstandings would also have influenced a different inspector.

A fifth intervention is landlord escalation. Repeated resident concerns, smoke-control defects, missing fire notices and incomplete vulnerable-person records could have triggered an independent whole-building review. Such a review might not have discovered the panel core without facade expertise and records, but it could have exposed weak management, incomplete strategy, missing barriers or emergency information. The inference is plausible, not certain.

A sixth intervention is earlier operational recognition on 14 June. Once external fire spread and multiple flat involvement were visible, an evacuation strategy and mass-call process designed in advance could have changed advice earlier and coordinated the stair. The Inquiry found LFB preparation deficient, supporting the conclusion that better preparation could have improved response. It cannot establish a minute-by-minute alternative casualty outcome because smoke conditions, individual locations and mobility varied.

The point of these counterfactuals is not to select a single substitute cause. It is to show that independent barriers existed at regulation, product evidence, design, construction, approval, management and response. A resilient system does not assume the first barrier will always work. It expects later controls to detect and stop an unsafe condition before residents bear the consequence.

A durable accountability test for building safety

The first test is control ownership. Can every safety-critical function be assigned to a named legal and operational owner: product claim, wall design, substitution, installation, inspection, building approval, fire-risk management, resident communication, evacuation and emergency information? Shared work is acceptable. Unowned decisions are not.

The second test is evidence integrity. Can a reviewer trace each claim from the installed product to a complete test, specimen and valid field of application? Are unfavorable results, modifications and uncertainty disclosed? Does certification challenge the supplier independently? A certificate without traceability is not assurance.

The third test is system thinking. Has the actual combination of panels, insulation, membranes, barriers, windows, fixings and cavities been assessed for credible fire scenarios? Are interfaces and substitutions controlled? Product compliance cannot be added together to prove wall compliance.

The fourth test is competence and independence. Are architects, fire engineers, contractors, facade specialists, assessors and building-control professionals qualified for the work they perform? Are commercial conflicts declared and managed? Can the reviewer stop work without depending on the party whose schedule or sale is threatened?

The fifth test is resident power. Can residents obtain safety information, report concerns, see how evidence was assessed and escalate beyond the landlord? Are vulnerability and evacuation arrangements developed with consent? Consultation that cannot trigger investigation is not accountability.

The sixth test is lifecycle information. Does the golden thread reflect the installed building, later changes, inspections, occurrences and remediation? Can an incoming owner, manager or fire service understand the risks without reconstructing decades of missing files? Information must be current enough to support action, not merely preserved.

The seventh test is emergency adaptability. Is stay put tied to verified compartmentation and monitored conditions? Can incident commanders change strategy, communicate it and support those unable to self-evacuate? Are control rooms and fireground systems tested for mass simultaneous calls and degraded communications?

The eighth test is remediation proof. Is each unsafe condition identified, funded, fixed and independently accepted? Do published counts distinguish work started, physically complete, awaiting sign-off and fully verified? Are non-cladding defects included where they remain life-critical? Programme enrollment is not completion.

The ninth test is enforcement. Can regulators detect noncompliance, obtain evidence, impose timely measures and prevent actors from continuing business while serious obligations remain unmet? Are enforcement outcomes and response times visible enough for public scrutiny? Rules without credible enforcement recreate the pre-Grenfell incentive to pass risk onward.

The tenth test is institutional learning. Are inquiry, coroner, incident, complaint and near-miss recommendations tracked to an owner, deadline, evidence of completion and independent review? The 2026 recommendations register is a useful mechanism. Its value will depend on whether "complete" means the intended safety outcome has been achieved and sustained.

Grenfell Tower was not an unknowable accident caused by one exotic failure. A domestic appliance fire encountered a wall that should not have been on a high-rise home, after a long sequence in which warning, test, design, procurement, approval and resident evidence failed to stop it. The emergency system then confronted conditions it had not prepared to manage. The Inquiry's account makes fragmentation visible without making it an excuse.

The repair programme has changed law, institutions, duties and thousands of facades. That progress matters. So do the 2,080 monitored buildings whose unsafe-cladding work had not started at the end of May 2026, the incomplete recommendation programme, the unresolved criminal process and the limited public access to building-level assurance. Institutional legitimacy will not be restored by saying the system has learned. It will be restored when residents can see who controls their building's risks, inspect the evidence that those risks were reduced, and obtain action before another warning becomes an inquiry finding.