The question, and the yardstick
The same question applies to this project as to the season’s other one, the Fidelbo Committee’s reform of Decree 231, because the two are children of the same hand: does a workplace-safety reform serve to reduce the dead and the injured, with the right safeguards in the right place — or does it serve to protect the top of the company even further from trial, without adding a gram of real prevention? Only here the question has a cruder yardstick than elsewhere, because the subject matter is not money: it is bodies. Before opening the report, then, one must count. Not out of emotion: out of method. A reform is measured against the problem it claims to solve, and the problem, here, has precise numbers — and numbers that are missing, which is worse.
Part one. Deaths at work. The number no longer scandalises anyone
The number in circulation is one: roughly three deaths at work per day. It is true, and it is already a selection. In 2025, the reports of fatal accidents received by INAIL, the national insurer for workplace accidents, numbered 1,093 — 792 “in the course of work”, 293 on the commute — out of 597,710 reported accidents; in the first five months of 2026, 370 fatal reports. But year-end reports mature: 2024, closed at 1,090, rose to 1,202 in the consolidated survey. And reports are not findings: of the reported dead, those INAIL establishes as having died “in the course of work” oscillate between 550 and 886. The public number of deaths at work is therefore an interval — between the five hundred and fifty established and the twelve hundred reported — which depends on who counts, when they count, and what they decide to count.
Two trends must be fixed, because the reform will have to answer for them. The first: deaths from accidents are not falling — setting aside the COVID years, fatal reports have oscillated since 2019 between 1,041 and 1,093, a flat line that INAIL itself describes as “in line with 2019”. The second: occupational diseases are exploding — from 45,000 reports in 2020 to 98,463 in 2025, the highest since the Seventies, up 119 per cent in five years, and 2026 is running towards a third consecutive record. Any safety reform written in 2026 is born inside these two facts: fatal accidents stagnant, occupational disease in vertical growth.
Hunting for what lies inside the 98,463 reports, and for what does not. The scandal of occupational diseases… and of PFAS
The count of “deaths at work” is built on accidents: the traumatic event, dated, visible. It is not built on exposures: the poisonous substance breathed for twenty years, the cancer that arrives thirty years later, the death that the certificate attributes to the disease and not to the factory. To see what is missing, the occupational-disease number must be taken apart — because, put like that, “98,463 reports” does not say enough.
Three questions open it up: what it contains; how much of it is recognised; what it does not contain at all.
What it contains. The archive is dominated by the diseases that can be seen and do not kill: in 2024, out of 88,499 reports, 74.7 per cent concerned the musculoskeletal system, another 12 per cent the nervous system (carpal tunnel, above all), 6.5 per cent noise-induced hearing loss. Backs, tendons, ears: more than nine reports out of ten. These are real diseases, and their growth speaks of work that consumes bodies and, in part, of a greater propensity to report. But the part that kills is a thin fraction: reports of occupational cancer run at about two thousand a year — two per cent. Yes, then: the archive also holds asbestos, vinyl chloride, the recognised carcinogens. The right question is to what extent they are there.
To what extent, the comparisons with science tell us. Asbestos: the National Institute of Health (Istituto Superiore di Sanità) estimates 4,400 deaths a year from asbestos-related diseases — 1,545 from mesothelioma, the rest mostly lung cancers —; INAIL, within the same perimeter, recognises about 1,160. A factor of four, for the country’s oldest, most studied and most regulated occupational disease: and if this is the fate of asbestos, which has tables, a national registry and fifty years of case law, imagine the rest. Cancers as a whole: the literature attributes to work between 4 and 8 per cent of cancers; on Italy’s 390,000 annual diagnoses, the prudent share gives about 17,000 cases a year of occupational origin. The reports are 2,000; the recognitions, fewer than 940. Here is the abyss, stated precisely: it does not run between two separate lists — it runs between the scientific estimate and the archive, and it runs through the archive itself. Of the 17,000 estimated cancers, 15,000 are not even reported: the doctor makes no connection, the decades-long latency confounds, the sick person does not know. Of the 2,000 reported, more than half are not recognised. On the record remains one in eighteen. The archive grows on the pathologies that do not kill; the part that kills enters by the dropper, and the rest is lost three times over — to the statistics, which do not count it; to the victim, who is not compensated; to prevention, which cannot prevent what it does not see.
And then there is the limit case, where the abyss shows itself in its pure state: PFAS, and PFOA above all. Ubiquitous — it is in the blood of nearly every population studied; eternal — it does not degrade: chemistry calls them forever chemicals; a certain human carcinogen — IARC Group 1 since December 2023, the same box as asbestos and benzene. It has been in the working environment for decades, and for decades it has been poisoning water, soil and air.
In the Italian statistics of occupational disease, PFOA does not exist. Not a dedicated table, not a historical series, not a mortality entry.
Here one sees the lag of an entire scientific and social culture, and it is frightening.
Where PFAS are considered and counted, the numbers exist.
In the United States, a lawyer representing tens of thousands of people has been suing DuPont since 1999; in 2001 DuPont gives way and begins to pay — the settlements will climb to hundreds of millions, then to billions of dollars —; an epidemiological study of about seventy thousand people establishes the probable link with six diseases, two cancers included. The lawyer’s courage becomes legend, the story becomes a film, “Dark Waters”. We watch it too, in Italy. The film. And when we hear the word PFOA we say: what on earth is this stuff? We don’t have it here, or we would know.
The Miteni investigation starts in 2018 — twenty years after America — and in Veneto they discover that their waters are far darker than the American ones.
In Italy the numbers do not exist: there is Miteni, and that is all. But Miteni had become a peripheral production unit of Solvay and DuPont; and yet the admittedly commendable Vicenza Public Prosecutor’s Office does not knock on the giants’ door.
In words, because the veil of corporate personality is sacred — while in America, for equal facts, Solvay and DuPont are made to hand over hyperbolic sums.
In fact, for political reasons: “careful, or foreign capital will stop coming”.
The first and only recognition of a PFAS death as an occupational disease dates from 13 May 2025 — Pasqualino Zenere, a worker at Trissino from 1979 to 1992, dead of urothelial carcinoma in 2014: for eleven years that death existed in no column.
And the University of Padua’s study with the National Institute of Health estimates, in the Veneto “red zone” alone, nearly four thousand excess deaths between 1985 and 2018: four thousand deaths that appear neither among the reports nor among the recognitions — for the archive, they never happened.
Epidemic-scale numbers, and what is more, numbers underestimated, and not by a little.
A certain carcinogen, everywhere and forever, with a single official death: this is the exact measure of what the national count does not see. The historical diagnosis is that we are a country that is a colony — in this too. And the silence is never innocent.
The Miteni case reveals the disconcerting idleness of all the public bodies which by law should watch over and protect the supreme collective good, the life and health of the citizens who entered the Social Contract.
If the State does not defend man’s primordial goods, what is it for?
Now the report can be opened, with a single simple question: how many of these deaths does the reform try to prevent?
Part two. Protecting the employer from the criminal law that should protect workers’ lives
The study Commission “for prevention and safety in the workplace” was established by decree of the Minister of Justice of 27 March 2024 — six weeks after the Fidelbo Committee, same Ministry — and is chaired by deputy minister Francesco Paolo Sisto. It convened on 24 April 2024, held thirty-eight meetings, closed its work on 22 July 2025; the report bears the date of 30 April 2026 and was delivered to the Minister on 12 May — between closure and delivery, nearly ten unexplained months. The product is a set of draft articles in three parts: the Criminal Code, the procedural code, the Consolidated Safety Act. The intended outlet is an enabling law. As of 31 July 2026, no government bill appears to have been tabled.
Who wrote it: twelve members besides the chairman — two full professors of criminal law plus an associate professor seconded to the Legislative Office, the same one who sits on the Fidelbo Committee: the two projects share a member; a criminal judge of the Court of Cassation, two criminal-defence lawyers, the deputy head of the Minister’s private office, the president of INAIL, two academic engineers, a full professor of occupational medicine, an environment-and-safety officer of a territorial Confindustria association. The knowledge summoned: criminal law, engineering, the public insurer, business.
The list of who is not there is more eloquent. No workers’ representative. No victims’ association. No labour judge. No inspector: neither the National Labour Inspectorate nor the prevention services of the local health authorities — the two public structures that actually see workplaces. In the commission that rewrites criminal liability for deaths at work, the party of the dead does not sit.
The hearings confirm it: INAIL’s statistics office, two judges of the Court of Cassation, a single labour-law scholar, a senior adviser of Confindustria. Not heard: CGIL, CISL and UIL; not ANMIL and the victims; not the Inspectorate; not the technicians of the local health authorities. Here the criminal protection of workers’ lives is being rewritten without listening either to the workers or to those who collect their bodies; and Confindustria is inside twice — heard as an expert, present as a member.
As for the empirical basis: a single study in support of the rewarding mechanism — the Accredia study, according to which companies with certified systems show an accident frequency more than 20 per cent lower. Accredia is the body that accredits the certifiers of the very systems the study evaluates: it is not a third-party source; and it is a correlation, not a cause — those who get certified are already, at the outset, better organised. On this rests the entire rewarding structure. No study on the effect of the models on accidents; no analysis of the judgments; no data on occupational diseases. The yardstick of part one, in the report, does not appear.
Part three. The safe-conduct — not for the workers, but for the employer
The heart is the new Art. 590-septies of the Criminal Code. Paragraph 1, verbatim: “Where an adequate organisation and management model under Art. 30 of Legislative Decree no. 81 of 9 April 2008 […] has been adopted, the employer […] is punishable for the offences under the second paragraph of Art. 589 and the third paragraph of Art. 590 only in the event of gross negligence”. Translated: if the entity has adopted an “adequate” model,
the employer answers for negligent homicide and gravest bodily harm with breach of accident-prevention rules only in the event of gross negligence. The employer’s ordinary negligence — the carelessness, imprudence or incompetence that caused death or injury — becomes criminally irrelevant. And the benefit applies to the employer alone: not to the managers, not to the supervisors, not to the head of the prevention service.
What gross negligence is, the text does not define except by examples.
Paragraph 3 lists the breaches that make the negligence “grave in any event”: failure to appoint the occupational physician or the head of the prevention service, failure to assess the risks, failure to draw up the document, failure to adopt the measures, failure to update, failure to supply the protective equipment, failure to inform and train.
The total omissions — having done nothing at all: cases infrequent, to put it mildly, if any still exist. The class of the “grave in any event” is drawn on an almost empty set. Everything above it — the assessment done badly, the measure chosen badly, the risk underestimated, the maintenance skimped, the control slackened — enters the territory where the judge will have to ask whether the negligence was “gross”, with the adopted and perhaps attested model arguing for no.
Paragraph 2 lists the indices the judge “takes into account”: the nature and complexity of the activity; the knowledge of the risk — and here the question already presses: the knowledge actually held or the knowledge owed? The text does not say —; the good practices “validated by the standing Consultative Commission”; and, fourth index, possession of the attestation of the joint body “or” of the certification of the management system “issued by an accredited body”.
A datum of experience, declared as testimony: in decades of workplace criminal trials we have never encountered either one in a case file. Neither attestation nor certification. They are rare objects in the places where people die; and yet it is to these objects that the project entrusts the measure of negligence.
We foresee hordes of certifiers and verifiers, and the market that will be born will be at least the equal of the 231 market.
Supposing an “adequate” model exists, is each and every negligence of the employer forgiven?
If that were so, the provision would go straight to the Constitutional Court.
The clause of the eventual new 590-septies seems, a contrario, to establish a curious equivalence between the inadequacy of the model, that is, the breach of that precautionary rule, and negligence, that is, the breach of any other precautionary rule whatsoever.
An “adequate” model but a free fall in maintenance and safety spending? If you have seen to the adequate model, you are no longer negligent, full stop! Save for gross negligence, which is the pneumatic void described above.
The finale.
What stays out — of this article and of the entire project — is the true fault of the top: the company’s safety policy, the one that is read not in the abstract model but in daily practice, or in the accounts — in the line items on maintenance costs and investment, in the cuts made and their reasons, and so on.
It is the fault that this country’s gravest trials have always found there, in the numbers, not in the papers of conformity.
On that, the project has not a word.
More on the double-faced impunity bought in a shop
A few further notes, because the reform under way will be hailed as a beneficial liberalising reform — about time, at last.
Art. 30, paragraph 5, of the Consolidated Safety Act reads word for word: “On first application, corporate organisation models defined in conformity with the UNI-INAIL Guidelines for an occupational health and safety management system (SGSL) of 28 September 2001 or with British Standard OHSAS 18001:2007 are presumed to conform to the requirements of this article for the corresponding parts”.
Read it again slowly. The model conforming to X is presumed to conform to Art. 30 — where X is a guideline of 2001 or a private British standard of 2007. The “first application” has lasted since 2008, but the government bill of 4 August is preparing to replace them with UNI ISO 45001 certification.
X is not fixed by the judge, nor written by Parliament: X sits in the window of the shop that the legislator recommends — the standards body, the accredited certifier, the joint body that attests.
The farce nests exactly here. From this point on, one need only follow the chain and find the four links, each written in a different provision, none of them secret.
First link, the presumption: the model conforming to X is presumed to conform to the requirements of Art. 30 — paragraph 5, just read. Second link, the company not punished: the Art. 30 model is, by definition of its paragraph 1, the one “suitable to have exempting effect from the administrative liability of legal persons” under Decree 231 — so the company, with the model conforming to X, is exempt. Third link, the employer not punished: the proposed 590-septies hooks the natural person’s shield to that same model. Fourth link, the valve: gross negligence would remain — but the reform, in paragraph 3, as we have seen, declares “grave in any event” only having done nothing at all; and whoever bought the model cannot, by definition, fall within paragraph 3.
A presumption of abstract adequacy of a document called a model — abstract rules, chosen — supports the double non-punishability: Decree 231 and the offence committed with non-gross negligence, which is nearly always.
Pay for one, get two.
Now weld the links and read the chain in full, because it is the provision the project does not have the courage to write in a single article: model conforming to X, employer and company not punishable, even if the employer breached another precautionary rule — as long as it is not gross negligence; and gross negligence exists only for those who did not buy the model and did nothing at all, displaying the rarest idiocy (doing a little something is enough…).
Impunity is purchased in a shop, and the window was dressed by the legislator.
Who keeps the shop, and at what textual price?
The parties who attest conformity to X are chosen and paid by the employer; and paragraph 2 orders the judge to “take account” precisely of their stamps: whoever sells the model sells, in the same package, the proof of its goodness.
And the words must be weighed to the end.
What is the legal nature of this impunity?
Here is the device’s refinement: each link, taken alone, is a rebuttable presumption, defeatable; the whole chain, in fact, is not — because to defeat it the judge would have to declare inadequate a model that a statutory provision presumes conforming, that a certifier has certified, that a joint body has attested, and of whose stamps another provision orders him to take account. A presumption of fact which, under the right conditions — the right vendor, the right stamp —, works like a presumption iuris et de iure. No provision declares it; the ensemble produces it.
And measure what is being given away to the company: what is Decree 231 liability for inadequacy of the model worth in the trials?
In the trial for the gravest water poisoning in the history of the Republic — the Miteni of which we wrote above — the entity answered, under Decree 231, with 125,000 euros, against the 141 years imposed on the men. That was already the light pan of the scales; the mechanism takes even that off the scales.
The reverse confirms the obverse. For those who do not buy the model — the micro-enterprise, typically — everything stays as today, indeed worse: ordinary negligence and increased sentencing ranges.
The reform does not raise the threshold of punishability: it pegs it to organisational wealth. Above the shop’s threshold, impunity save for total omission; below it, heavier penalties.
And since a criminal system, to call itself one, needs someone left to accuse, look finally at who remains.
The shield covers the employer alone; the adequate model covers the entity. Managers and supervisors remain punishable for ordinary negligence — with the increased penalties. As one descends the steps of corporate power, the criminally sanctioned duties climb.
The head of the prevention service acquires liability for offences of his own, with the new Art. 58-bis.
The final map: at the top, no one; at the bottom, everyone (potentially).
The cascade of delegations and sub-delegations will slide down the slope of the organisation chart until it finds the accusable one: the manager, the supervisor, the technician — the one who signed the last paper without holding the first euro.
The scapegoat is not a manufacturing defect of the device: it is its end product. The law does not abolish criminal liability for death at work; it redistributes it by organisational wealth — switches it off at the top, deposits it where the delegations end.
The historic slide: Decree 231 was born as a new liability of the entity, alongside the employer’s. Now Decree 231, with its model presumed “adequate” and notarised with a stamp, safeguards itself and renders the other irresponsible: the employer
Decree 231 was born, in 2001, to break a dogma that had crossed two centuries: societas delinquere non potest — only man answers, because only man has will and fault. The rupture served to add a responsible party, not to substitute one: alongside the natural person, from then on, the organisation that made the offence possible would answer too; and the Decree carved it into Art. 8 — the entity’s liability is autonomous, it subsists even when the author of the offence is not identified. Two liabilities, side by side, so that the organisation would no longer be the screen behind which the man, the employer, vanished.
In twenty-five years, among all the distortions documented by this series, that architectural principle had never been touched: never had Decree 231 been used as a surrogate for the criminal liability of the person. It was born, literally, for the opposite.
The Sisto project makes it come full circle in a single article: the entity’s model screens the man, and meanwhile goes on exempting the entity.
The instrument created so that the organisation would no longer serve as a screen becomes a screen twice over. If one wants the Latin that tradition demands, the tenses are three.
Societas delinquere non potest: the nineteenth century — only the man paid. Societas delinquere et puniri potest: 2001 — both were to pay. The project writes the third tense, and it resembles neither of the first two: absolvitur societas, absolvitur dominus — manet caper emissarius.
The company absolved, the master absolved: the scapegoat remains. Which is, from Leviticus onwards, the animal loaded with faults decided by others — here renamed: the Delegate.
The project rereads Art. 27 of the Constitution — “criminal liability is personal”, written as a guarantee: no one punished for another’s deed — and makes it live now as a privilege: the employer will not be punished for his own deed, provided the entity has adopted the right paper. The guarantee was written to protect man from the State; the privilege protects the top from the trial that should protect the victims. They are the same sentence, read from two opposite sides of the table — and the project has chosen its side.
The system, to this day, proclaims that the employer is the person who more than any other — artificial or not — guarantees the life, safety and health of the workers.
The Constitution is still interpreted by everyone as the insuperable source of the primacy of health and safety at work, ranked above any other value — above all the economic one, with which it conflicts.
Duly noted, then.
The assessment of risks to workers
The risk assessment remains a non-delegable duty of the employer (Art. 17 is untouched), but look at the paradox: the employer answers if the document does not exist; if it exists and is mere paper, he no longer answers for it.
In the new Art. 33, it is no longer “the service” that carries out the assessment but “the head of the service”, “by reason of his technical competences”: the function passes from the collective organ to the person.
The new Art. 31 keeps the employer’s duty non-derogable, with a ban on delegation to whoever will in any case be liable for it — and two paragraphs close the design.
Paragraph 7: the employer’s duty of supervision “is deemed discharged where the verification and control model under Article 30, paragraph 4, has been adopted and effectively implemented”.
Paragraph 8, verbatim: “The employer does not answer for the inadequacy of the risk assessment if he has correctly discharged the obligations laid down in this article”.
Whoever appoints, equips and funds the service does not answer for the quality of the assessment the service produces.
A new Art. 58-bis introduces offences proper to the head of the service — detention from three to six months or a fine of up to a little over nine thousand euros, according to the concurring readings of the first commentaries.
The figure whom the case law considers devoid of autonomous criminal liability becomes a centre of imputation.
The chain must be read and reread: the duty stays in the employer’s name, the content is written by his employee or consultant, the error in the content is no longer imputable to the employer, and a new available culprit has been born — internal, technical, insurable.
The technical professions understood it before anyone: they speak of a “lightning rod” and of a “lame guarantor” — responsible for everything, master of nothing, because the spending powers stay where they have always been. At the top switched off, in the middle no one, at the bottom the technician: the relocation is the descent already described, put into law.
Part four. And the State? It grieves, protests, pledges, then throws in the towel with great dignity
In the current system, the risk assessment, as written, is signed by the employer, with technicians who materially produce it for him and are paid by him, and no public authority sees it before something happens — the document is not filed, not validated, not checked ex ante; public control is the sample inspection, or the trial, after the death.
The conflict of interest is structural: whoever bears the cost of prevention decides how much prevention is needed. In any other field where life is at stake — the drug, the aeroplane, the bridge — the producer’s self-assessment meets an independent public eye; in work, it does not.
The reform could choose its direction, and it chose: no ex ante control even now, no filing, no strengthening of inspection — only the three internal relocations already seen.
It is not the privatisation of health protection: that already existed. It is its perfection — privatisation with indemnity.
And the numbers of public control are the scandal everyone knows, including Sisto’s reformers, yet nothing is said of it, so that nothing need be done.
Active enterprises in Italy number 4,617,000; the National Labour Inspectorate has 2,969 civilian inspectors, of whom 949 are technical. The derivation — a calculation, declared as such — gives an annual probability of inspection of 3.4 per cent, once every twenty-nine years; isolating health-and-safety supervision, 1.1 per cent — once every eighty-nine years.
The irregularity rate among inspected enterprises has stood between 70 and 74 per cent for seven years: one goes where one knows one will find, and three out of four are irregular.
Meanwhile the “validated good practices”, to which the new negligence index refers, have stood still since 27 November 2013: for thirteen years the public body in charge has validated none.
The comparison with medical negligence aggravates rather than attenuates: health-care guidelines are drawn up by scientific societies entered in a ministerial register and published by the National Institute of Health; the attestation and the certification are produced by the parties and by private operators, without the project dictating a single requirement for who attests and who certifies.
In Italy, public control is a farce performed, however, always on the company’s side and never on labour’s.
To the theme — which in US doctrine bears the name of a criminological class of offences: State-facilitated crimes — we will devote a separate investigation, because in Italy the opposite principle reigns: the State and its local articulations not only do not act and do not answer, but punctually join the proceedings as civil parties against the accused company.
For show, naturally.
And Europe? And what do we in Italy make of the obligations Europe imposes on us?
There is also an open account with Europe, and someone will take it to Luxembourg: framework Directive 89/391 makes the employer responsible for safety “in all aspects related to the work”.
A national law that exonerates him of all negligence for the mere paper adequacy of the model and of the assessment has, with that directive, a compatibility problem which the technical associations themselves denounce.
The Decree 231 penalties? The light pan of the Miteni scales is destined to stay light by choice, not by chance.
Europe has just ordered it to be weighed differently: Directive (EU) 2024/1203 on the criminal protection of the environment requires, for legal persons, penalties with maxima of no less than 3 per cent of worldwide turnover — or 24 million euros — and, for the gravest offences, 5 per cent, or 40 million.
Italy transposed it with Legislative Decree no. 81 of 21 April 2026 — the same number, by irony of fate, as the Consolidated Safety Act —, in force since 2 June: and the ministerial report declares that it declined the turnover criterion, “for proportionality”, keeping the quota system.
Why, no one knows.
Environmental disaster rises from 900 to 1,200 quotas: at the quota’s maximum value, not even two million euros.
Forty million demanded by Europe, less than two granted by Italy; and three weeks after that decree, the Sisto report reaches the Minister with the exemption as its dowry. With one hand one transposes downwards, with the other one exempts.
Such benevolence towards business will cost Italy a procedure for breaching the European directive.
And the Sisto reform, for workplace matters, in fact lowers the Decree 231 penalty compared with today: not in the ranges, which it does not even touch, but in effect — once the conforming model is bought, the exemption is presumed, and the penalty drops to zero.
Heavier penalties for the delegates, impunity for the delegator. New mitigating factors where aggravating ones would have sat well
The reform retouches the sentencing ranges: negligent homicide with breach of accident-prevention rules goes from two-to-seven years to two-and-a-half-to-eight; the penalties for bodily harm rise too.
The structure must be seen whole, because it is a two-faced coin: the symbolic face — the penalties going up — shown to the public; the effective face — the pool of the punishable emptying out — spendable in court by those who have equipped themselves.
In parallel it introduces a new mitigating factor: where the event “is not the exclusive consequence of the action or omission of the culprit”, the penalty may be reduced by up to a third for whoever made “a contribution of minimal importance”.
A rap on the drafters’ knuckles — let them not take offence, or let them, as they please —: never, in rerum natura, is there a condition at once necessary and sufficient; never is an event the exclusive consequence of a single action.
A concurrence of causes exists always and in any case.
The Criminal Code of 1930, fascist but not uncultured, laid down Art. 41, which enshrines, repeating it, the granite principle of nineteenth-century criminal-law culture: the logical equivalence, for the criminal law, of all the necessary conditions of the event.
The derogation for the supervening cause capable on its own of preventing the event was meaningless, and everyone knows it.
The conditions, all logically necessary, are not weighed: there is not one worth more or less than another.
The reform, however, wants to change the cards on the table, inventing a condition that weighs less than the others; and the trick is meant to regulate, sliding down by gravity, the multifactorial nature of occupational diseases.
As written at the outset, the reform here moves on the terrain of the great scandal, that is, the gigantic underestimation of reports, and still more of their acceptance, for diseases from occupational exposure.
Instead of more regulation and greater responsibilities, the one is omitted and the others are elided, where they belong to the top, to the employer, who more than anyone must transform the company’s culture to orient it towards real prevention.
Here, though, there is worse: either one knows and reasons well, or one does not know and reasons badly.
Either the work was a necessary condition of the event, together with many others in concurrence — and then nulla quaestio: it counts like the others, with no discounts of any kind —; or it was not, and then it is not a discount that is needed: there is nothing to pay — employer and, down along the cascade of delegates, everyone acquitted, and that is that.
The reform betrays a deep decline of criminal-law culture in Italy
The concept, by now threadbare in the criminal-law lexicon, of “multifactorial disease” belongs to the epistemic domain of “risk”, not to that of the “necessary condition”.
There is a greater or lesser quantity of risk, not a greater or lesser quantity of causation.
The Italic reforming legislator, unfortunately, is ignorant both of the conditionalist paradigm, received from an august ancient tradition, showing he cannot handle it, and of the opposite paradigm of risk, which is made not of logic but of probability and much else.
It is the epochal point that measures whether a legal order has truly equipped itself for the passage, long since accomplished, to the Risk Society, or whether Ulrich Beck’s fortunate title is merely the ornament of the void that appears in the opening credits of publications that boast of being modern, indeed postmodern.
This botched mitigating factor is the quasi-legislator’s answer to the capital problem stated at the outset — the explosion of occupational disease — which is regulated here with a reduction of penalty that will have, we imagine, the general-preventive efficacy that was intended.
The passage, prudent and pondered, to risk is more than necessary; otherwise the living criminal law will meet other epistemic catastrophes such as the sad spectacle of prosecutors and judges grappling with the causal relevance of the brief periods of exposure imputable to the one and to the other defendants in succession of office over time.
But treating the probabilistic imputation of the event requires whole libraries of new studies: of logic, but also of the sociology of risk and, above all, of new sciences, such as epidemiology, which measures risks and does not establish conditions.
We at whitecollarcrimes.it will cross this passage to the end.
And the light-years lag of our criminal-law thought behind the extraordinary flowering of legal doctrines of risk in the USA will be touched with the hand.
The least bad page
The article on procedure contains the least bad page of the project, and it must be acknowledged: the “code red” fast-track model extended to workplace offences.
The public prosecutor must take information from the injured party within three days of registration; the criminal police execute delegated acts “without delay”; notice of a request for dismissal must always be served on the injured party, with the deadline for opposition extended from twenty to thirty days.
It is the exact opposite of the Fidelbo Committee’s choice, which excluded the victim’s opposition from its own dismissal: here the victim enters, there she has no citizenship. The two projects of the same Ministry, on the same figure, say opposite things — and neither notices.
The tragedy under way, or the guinea pigs and the absentees
The simplest question remains, dates in hand.
If reported occupational disease is growing as we have seen, what does the Sisto project propose as a remedy for the tragedy under way? Nothing. In the text there is no word “exposure”, no word “carcinogen”, no provision on health surveillance. But “nothing”, said in the abstract, cannot be measured.
To measure it, set beside the project the case that grew up next to it, in the same months, eighty kilometres from the ministerial tables: the PFAS case.
We treat it in outline, because this publication devotes a serial investigation to that story — “The Absent”, “The Guinea Pigs” — to which we refer for the whole. Here we need the three points that measure the legislator.
The first is work. The certain carcinogen of part one was breathed by the Trissino workers of Miteni as by the Montedison, Ausimont and now Solvay-Syensqo workers at Spinetta Marengo: PFOA blood loads among the highest ever documented in the world.
We called them “the guinea pigs”, and the word was chosen with care: imposing exposure of human bodies to a substance without consent, without protocol and — for the substitute molecules of the last decade — without even a study following over time the health of those who handled them, is exactly that.
The second is the trial. On 26 June 2025 the Vicenza Court of Assize closed the first instance — the convictions of the men and of the entity we have already weighed —; the two multinationals that had created the molecules that poisoned Veneto and governed the production cycles, Solvay and DuPont, never charged: “the absentees” — not acquitted: never judged.
The third is the legislator, and it must be entered in the register of absence — which in this publication means: the search was conducted, and it returned nothing.
In the United States, as seen, the litigation starts in 1998-2001 and ends in settlements running to billions, earmarked for compensation and clean-ups: up to 12.5 billion dollars from 3M, 1.19 from DuPont, Chemours and Corteva. And those papers were not unknown to the industry operating in Italy: the Vicenza judgment documents that already in 2005 Bilott’s letters were circulating in the e-mails of Solvay executives — they knew, for twenty years.
In Italy, over the same span of time: no national limit for PFAS in drinking water until 12 January 2026, when the European ones entered into force; the stricter limit on the four priority PFAS, PFOA included, and the monitoring of the molecules of the former Solvay plant at Spinetta Marengo postponed by six months by the latest budget law, “to give operators time to adapt”; no occupational exposure limit value in the Consolidated Safety Act; no provision easing collective civil actions by victims — seventy thousand claimants in a class action is the American number, fifteen thousand euros a head the Italian one —; no strengthening of the duties of public control, after the alarm over the new PFAS in Veneto had arrived, as the record shows, from a letter of the Dutch environment ministry and not from an Italian control.
And the Sisto project — the commission closes its work twenty-six days after the Assize judgment and two months after the Zenere judgment — contains not a line on any of this.
Over PFAS, something resembling a crusade is being fought in the world: on one side those who defend life and health — the epidemiologists, the occupational physicians, the lawyers like Bilott, the mothers of the red zone —; on the other, those who defend the company’s bottom line. Every legal order chooses its side, and chooses it with facts: with the limits it sets or postpones, with the controls it funds or leaves uncovered, with the actions it eases or abandons to individuals. The Italian legislator has chosen without declaring it. Before four thousand estimated deaths, silence is not neutrality: it is the taking of a side — the side of those who cause the death, not of those who suffer it.
The reactions, and the usual geometry
The documented reactions draw the same map as the parallel project. Against: the CGIL — “the reform that lightens companies’ responsibilities” —, the UIL — demanding “more prevention, training, inspectors and controls”, that is, exactly what is not in the text —, the opposition parties — the “criminal shield”.
Against, and it weighs more because it comes from inside the technical system: AIAS, which denounces the conflict with Directive 89/391; the National Council of Engineers, for whom “the employer’s responsibility must remain the primary lever of the system”; the technicians’ associations, which refuse the role of scapegoat.
In favour: the compliance sector, which speaks of an “operational turning point of great significance”.
In silence: ANMIL and the victims’ associations — the silence of the other side, which is not consent but absence from the field — and Confindustria, which has no need to speak: it was inside.
A parliamentary note, for completeness: before the Chamber of Deputies lies a proposal of the opposite sign — the autonomous offence of “homicide at work”, with penalties of up to twelve years for the gravely non-compliant employer.
The two texts photograph the two possible directions; only one of the two has a ministry behind it.
Part five. Judgment
On the yardstick set at the outset, the answer is clear: this project is not a reform of workplace safety; it is a staggering retreat of the criminal liability of the employer and of the company for death and disease at work.
Take up the numbers of part one, one by one.
To the 550-1,200 deaths a year from accidents, stagnant for five years, the project answers with an ex post reallocation of liability: nothing increases the probability that the risk will be removed beforehand — no additional control, no additional inspector, no new duty to act, only new rewarding effects for having documented.
To the 98,463 reports of occupational disease it answers with silence: not one provision on exposures, carcinogens, asbestos — for the share of disease that is not even in the statistics.
To the eight-to-ten thousand estimated deaths from exposure no one answers, because no one brought them to the table.
To the three million irregular workers, invisible to any organisational model, it does not devote a line. Nothing on subcontracting; nothing on the licence-by-points; nothing on Art. 25-septies of Decree 231: the coordination with the Fidelbo Committee, evoked in the presentations, does not exist in the text — the parallel project is never cited.
The reform is calibrated on the only segment of the problem that coincides with the criminal trial that matters to it: the segment in which the problem is the company’s judicial exposure, not the worker’s exposure to risk.
The risk is assessed as before (alone), controlled as before (almost never), inspected as before (with the same staffing); what changes is what happens after the death.
Protection from trial grows; protection from risk stays identical.
With three true exceptions — the code red, the staffing of the service, the sentencing ranges — real, minority, and politically the most fragile, because they have not a single organised defender; conditional gross negligence has them all.
There is an internal counter-proof, and it is the bitterest.
The report justifies gross negligence with the “impulse to disengagement”: punishing everything discourages doing. One is tempted to laugh. Where is the evidence that in Italy everything is punished in workplace matters?
The symmetrical argument is never put: in what way does a regime deter crime in which, once the model is bought, ordinary negligence is no longer punishable, even beyond the model, and the inadequacy of the assessment is no longer the employer’s responsibility, except in the opposite florilegia of words?
If indiscriminate punishment produces disengagement — and it is not true, even supposing it really existed —, impunity conditional on documentation produces documentation. The Accredia study — the only number brought in support — cannot tell the two things apart. On this unresolved ambiguity rests the project’s entire preventive wager.
The weld with the parallel project need be said only once: the same document — the Art. 30 model, bought on the conformity market, attested and certified by private parties paid by the party verified — exempts the entity, opens the extinction of the offence in the twin project, is presumed conforming, and now screens the top.
One document, four effects. Two commissions of the same Ministry, one member in common, no mutual citation, a single beneficiary. If both became law, the Italian legal order would have built, for death at work, the most complete franchise in Europe: and it would have built it calling it prevention.
Something else is needed, and it can be said in a few lines.
Bring the submerged into the count: a national system for registering exposures and work-related diseases, with public data — what is not counted is not prevented, and the Zenere case — the Miteni worker dead of PFOA — shows what it means to entrust the count to the tenacity of the families.
On exposures, what the PFAS case indicates to the letter: limits set and not postponed, an occupational limit value for the substances the evidence indicates, biomonitoring of the exposed for as long as they live, collective procedural instruments for the victims.
A public eye inside the self-assessment: the filing of the assessment document, targeted control of high-risk activities, third-party validation for the processes that kill.
The inspectors before any other provision: every reform of liability that does not touch the probability of control is a reform of the probability of impunity. If gross negligence is wanted, define it in statute by magnitudes of governed risk and unhook its indices from the papers on sale. Give the head of the service, together with the offences proper to him, the powers proper to him: whoever answers must be able to decide and to spend — otherwise the scapegoat has been written into the law. And recompose the projects: a single reform, a single parliamentary examination, with inside it the parties missing from the tables — because the original defect of both projects, before any article, is the list of the absent.
The political risk, in one line: the project will enter circulation as “the reform that raises the penalties for deaths at work” — the symbolic face of the coin — and the effective face will travel in the technical reports.
But the underlying judgment can be said in a few lines, and with them we close.
The reform has a single true movement: the slide. At the top, liability does not move: it is switched off — the same sheet of paper shields the employer and exempts the entity, having it both ways with a single stamp.
Then everything descends: from the top to the technician with his proper offences, from the trial to the paper, from the deed to the document, from judgment to the purchased attestation — down, delegation after delegation, to the scapegoat.
Everything slides, except one thing: the risk.
That stays exactly where it has always been — in the air of the factory, in the water of the aquifer, in the blood of those who work.
Absolvitur societas, absolvitur dominus: manet caper emissarius. Et manet etiam mors: the death of the workers remains — as before, or perhaps more than before, because impunity bought with paper produces paper, not prevention. The guinea pigs do not slide. They are still waiting for someone to count them.
Note on sources
Principal document: Final report of the study Commission for prevention and safety in the workplace (chair: F.P. Sisto), dated 30 April 2026, delivered to the Minister of Justice on 12 May 2026, with draft articles in three parts; published by Giurisprudenza Penale (21 May 2026) and Sistema Penale (18 May 2026). Quotations in inverted commas are transcribed from the text; the closing provisions of the draft articles (Arts. 34 and 55 of the Consolidated Act and the new Art. 58-bis) are reported according to the report’s own description and, for the sentencing ranges, according to the concurring readings of the first commentaries. Data: INAIL, open data 2023-2026 and annual reports; composition of the 2024 reports (88,499: musculoskeletal 74.7 per cent, nervous system 12, hearing loss 6.5) from “Dati INAIL” no. 1, January 2025; ISS, Rapporto Istisan 24|18; ReNaM, VIII Report (2025); calculations on AIRTUM/ISTAT for occupational cancers; ILO (2023) and EU-OSHA estimates; ISTAT, Non-observed economy (2025); Eurostat, Accidents at work; National Labour Inspectorate, reports 2021-2025; ministerial pages on validated good practices (most recent: 27 November 2013); Prime Ministerial Decree of 17 December 2007. On the device and the great slide: Legislative Decree 231/2001, in particular Arts. 8 (autonomy of the entity’s liability) and 25-septies; Art. 30, paragraph 1, of Legislative Decree 81/2008, verbatim: the organisation and management model “suitable to have exempting effect from the administrative liability of legal persons” under Decree 231/2001 “must be adopted and effectively implemented”; Art. 30, paragraph 5, verbatim: “On first application, corporate organisation models defined in conformity with the UNI-INAIL Guidelines for an occupational health and safety management system (SGSL) of 28 September 2001 or with British Standard OHSAS 18001:2007 are presumed to conform to the requirements of this article for the corresponding parts”; Art. 16, paragraph 3, of Legislative Decree 81/2008 (“The obligation referred to in the first sentence is deemed discharged where the verification and control model under Article 30, paragraph 4, has been adopted and effectively implemented”); Art. 41 of the Criminal Code (concurrence of causes: equivalence of the necessary conditions); Art. 27, paragraph 1, of the Constitution. On the European conflict: Directive (EU) 2024/1203 on the criminal protection of the environment (penalties for legal persons with maxima of no less than 3 per cent of total worldwide turnover or 24 million euros and, for the gravest offences, 5 per cent or 40 million); implemented by Legislative Decree no. 81 of 21 April 2026 (Official Gazette no. 113 of 18 May 2026, in force since 2 June 2026), which keeps the quota system, declaring in the ministerial report that it does not adopt the turnover criterion; environmental disaster from 900 to 1,200 quotas (maximum quota value: 1,549 euros, Art. 10 of Legislative Decree 231/2001). The formulas of the proposed Art. 590-septies are reported according to the reproductions of the draft articles published by the journals and the first commentaries (Sistema Penale, 18 May 2026; Giurisprudenza Penale, 21 May 2026; RPLT; Lavorosì; PuntoSicuro), to be checked against the official text before any use in court documents. On the PFAS case: judgment of the Vicenza Court of Assize no. 1/25, delivered on 26 June 2025 (first instance; 2,062 pages), cited by page in our investigation “The Absent” (19 June 2026), from which come the operative part (141 years; the entity under Art. 25-undecies: 125,000 euros and confiscation of 437,500), the 2005 e-mails with Bilott’s letters among Solvay executives and the Dutch ministry’s letter of March 2018; the scientific review is in our “The Guinea Pigs” (part two, July 2026), with the primary sources filed there — IARC, Monographs vol. 135 (December 2023, PFOA Group 1); Girardi & Merler, Environmental Research 2019 (Trissino cohort: 462 workers, mean ~4,048 ng/mL, max 91,900); NASEM 2022; EFSA 2020. Vicenza Court, judgment of 13 May 2025, labour judge (Zenere case: urothelial carcinoma of the renal pelvis, link established “with high probability”, INAIL ordered to pay; press: lavialibera, Greenpeace Italia, Il Domani). Excess deaths in the red zone: Biggeri et al., Environmental Health, 2024 (University of Padua and ISS; ~3,900 excess deaths, 1985-2018, 30 municipalities). United States litigation: the Bilott-DuPont story (1998-2001) and the Leach class action; C8 Science Panel (about 70,000 participants; probable link with six diseases, including kidney and testicular cancer); settlements: DuPont 671 million dollars (2017); Chemours-DuPont-Corteva 1.185 billion dollars (2023); 3M up to 12.5 billion dollars (2023, public waterworks; releases and press: NPR, C&EN, Business Wire). Limits: Legislative Decree 18/2023 (Directive (EU) 2020/2184), in force since 12 January 2026 (sum of 20 PFAS: 100 ng/l; total PFAS: 500 ng/l); six-month postponement of the 20 ng/l limit for PFOA, PFOS, PFNA and PFHxS and of the monitoring of the Adv molecules of the former Solvay plant: 2026 budget law, Art. 1, paragraphs 622-623 (sources: il Salvagente, Greenpeace Italia). Directive 89/391/EEC, Art. 5. Reactions: Collettiva/CGIL (9 June 2026), il manifesto (30 June 2026), AIAS, CNI, technical associations, official presentations on GNews and Sistema Penale. The derived calculations (work-related deaths; annual probability of inspection), the register of absence (search conducted with negative outcome) and the single professional testimony are declared as such in the text. This article takes up and condenses, for publication, the extended analyses of the series “Enterprise versus Man”: “The Workplace-Safety Reform. The Sisto Commission’s Project” (31 July-2 August 2026), “The Purchased Rule” (8 August 2026), “The Absent” (19 June 2026) and “The Guinea Pigs” (July 2026). Any reported error will be checked and, if founded, corrected with due acknowledgement.
whitecollarcrimes
a collective byline · guarantor: Avv. Luca Santa Maria · 19 August 2026