With the implementation of the Occupational Safety and Health Disclosure System on August 1, 2026, an era has opened that mandates the public disclosure of corporate safety levels. In this new phase of industrial safety, which shifts from punishment-driven to proof-driven, the triangular structure of leadership, culture, and measurement dictates success or failure.
An Era Where Safety is 'Disclosed' Has Arrived
On August 1, 2026, the 'Occupational Safety and Health Status Disclosure' (Occupational Safety and Health Disclosure System), grounded in Article 10-2 of the Occupational Safety and Health Act, went into effect. As a follow-up measure to the amended Occupational Safety and Health Act passed by the National Assembly plenary session on January 29, 2026, and promulgated on February 19, the revision of the Enforcement Decree was deliberated and resolved at the Cabinet meeting on July 21, finalizing the subject entities and disclosure items. The entities subject to disclosure obligations are business owners employing 500 or more regular employees, construction business owners with an annual construction project value of 120 billion won or more, and public institutions, local corporations, and local public enterprises under the Act on the Management of Public Institutions.
The disclosure contents consist of statutory basic items and detailed items under the Enforcement Decree. The law stipulates the disclosure of the current status of the establishment of an occupational safety and health management system, industrial accident occurrences, safety and health performance from the previous year, safety and health activity plans for the current year, safety and health-related investment status, and industrial accident recurrence prevention measures and implementation plans. The Enforcement Decree added detailed items covering fatal accident statuses of affiliated supplier workers and fatal accident statuses of public construction ordering projects. The core weight of this system lies in the fact that accidents involving subcontractors and public procurement sites are also recorded on the report cards of primary contractors and ordering entities. Actual disclosures are made annually by April 30 through an internet website designated by the Minister of Employment and Labor, with the first regular disclosure to be fulfilled by April 30, 2027. Failure to disclose will result in the imposition of fines.
The significance of this system goes beyond simple information disclosure. Previously, a company's safety level was revealed only after an accident occurred, and even then, only fragmentarily through post-event procedures such as investigations and trials. Now, companies must present their safety management levels every year in numbers and plans, even in the absence of accidents. The design intent of the system is to move safety from a 'subject of punishment' to a 'subject of evaluation,' and further to a factor compared in the market. This change itself serves as a practical reason to discuss safety leadership and safety culture, because disclosure is ultimately a mirror revealing what an organization routinely does.
Numbers Have Improved, But the Structure Remains the Same
The indicators for the first half of 2026 are superficially encouraging. According to the "Supplementary Statistics on Industrial Accidents for the First Half of 2026" released by the Ministry of Employment and Labor in July, accident fatalities subject to accident investigation in the first half stood at 253 (232 cases), a decrease of 34 (11.8%) compared to 287 (287 cases) in the same period last year. This is the lowest figure for the first half since relevant statistics began in 2022, and the margin of decrease was the largest. By industry, the construction sector recorded 105 fatalities (103 cases), down 33 (23.9%), and other industries recorded 56 (54 cases), down 26 (31.7%).
However, the same statistics also send opposing signals. Accident fatalities in the manufacturing sector rose by 25 (37.3%) to 92 (75 cases). The Ministry of Employment and Labor analyzed the March fire at a Daejeon auto parts plant and the June explosion at a Daejeon defense company as major contributing factors to the increase. Accident types also showed divergent trends. According to the attached statistical tables of the same supplementary statistics, 'falls,' which account for the largest proportion of fatal accidents, decreased by 45 (34.9%) to 84 cases, driving the downward trend. In contrast, fires and explosions increased by 100% to 32 cases, and crushes and overturns increased by 88.9% to 34 cases.
This contrast suggests that control methods differ by accident type. While fall accidents show relatively direct effects from physical controls such as fall-prevention equipment and working platforms, fire and explosion accidents are likely driven by multiple management factors acting together, including process safety, management of change, and emergency response. Of course, falls are also strongly influenced by organizational factors such as work planning, supervision, time pressure, and subcontracting structures, and fires and explosions often rely heavily on equipment and engineering controls. Nevertheless, the trend shown by the indicators in the first half points to a stage where it is no longer possible to reduce the total number of fatalities simply by intensively managing a single hazard factor.
When interpreting statistics, two standards must be clearly distinguished. The figures above are based on 'accident fatalities subject to accident investigation,' while the "Industrial Accident Status," based on the Korea Workers' Compensation and Welfare Service's approval of industrial accident compensation, is separate. The 2025 statistics based on industrial accident compensation recorded 147,130 injured workers, 872 occupational accident fatalities, and 1,376 disease fatalities. Among accident fatalities, the construction sector accounted for 361 (41.4%), workplaces with fewer than 5 employees accounted for 354 (40.6%), workers aged 60 and over accounted for 450 (51.6%), and falls accounted for 280 (32.1%). Because the two statistics have different aggregation purposes and timing, direct comparison is impossible, and mixing them in corporate disclosures or ESG reporting compromises reliability in itself.
Why Punishment Failed to Build Culture
Four and a half years have passed since the Severe Accidents Punishment Act took effect on January 27, 2022. The design intent of the law was clear. By imposing the obligation to establish an occupational safety and health management system on the management executive holding final decision-making power over budgets and human resources, rather than on on-site managers, safety would be placed at the very top of corporate decision-making structures.
What were the points repeatedly contested in court judgments? A report titled "Status of Severe Accidents Punishment Act Judgments and Implications," published in January 2025 by the Korea Enterprises Federation, a business employers' organization, analyzed 31 cases (29 convictions, 2 acquittals) that received first-instance rulings as of the end of 2024 by judgment unit. According to the report, among the 29 convictions, the violation provisions most frequently cited by courts were Article 4, Paragraph 3 of the Enforcement Decree, 'preparation of procedures for identifying and improving hazard and risk factors' (24 cases), and Article 4, Paragraph 5, 'preparation of performance evaluation criteria for occupational safety and health managers, etc.' (22 cases), with an average of 3.07 violation provisions per judgment. While it must be taken into account that this report was prepared from a management perspective to raise the necessity of legal amendments, the citation frequency per clause itself shows where companies repeatedly tripped up. It was not high-difficulty technical measures, but basic management systems such as procedures to find and fix hazards and criteria for evaluating responsible personnel.
Skepticism regarding deterrent effects has also been raised from the perspective of sentencing. According to a legislative impact analysis report by the National Research Service for the National Assembly (as of July 2025), out of 49 cases resulting in convictions, 42 ended in suspended sentences, representing a rate of 85.7%. In the same report, out of 56 individuals receiving first-instance rulings, 6 were acquitted, recording an acquittal rate of 10.7%. Separately, a compilation reported by Weekly Hankook in June 2026 showed that 61 out of 70 natural person defendants (87.14%) received suspended sentences, though this cannot be directly compared with the aforementioned figures due to differing baseline dates and populations. However, both datasets align in the aspect that prison sentences are exceptional.
A significant portion of acquittals consists of cases where the causal relationship between the management executive's breach of duty and the fatality was negated. This creates an incentive for legal risk management to drift into criminal defense strategies. The fact that punishment can instill fear in an organization but fails to cultivate a culture of self-disclosure and improvement of hazards is what the judgments and statistics accumulated over four and a half years jointly suggest.
The Substance of Safety Leadership is Resource Allocation, Not Declarations
The concept of safety culture was first officially formalized through an International Atomic Energy Agency report following the 1986 Chernobyl nuclear accident. Research accumulated over the subsequent 40 years converges on a single conclusion: the biggest variable determining safety performance is not the amount of facility investment, but the actual signals sent by leaders regarding safety. Ever since organizational psychologist Dov Zohar empirically introduced the concept of 'safety climate' in 1980, it has been repeatedly confirmed that members read an organization's true priorities from the actions of management, not their words.
Practically speaking, safety leadership reduces to four observable levers. The first is resource allocation. Whether safety budgets are the first items cut during economic fluctuations or whether they are protected determines the organization's actual priorities. The second is time allocation. How many hours a top executive spends annually on safety-related meetings and how many times they visit the site dictates the attention distribution of the entire organization. The third is the direction of attention. An organization where leaders ask only about accident counts versus one that asks about near-miss report counts generate entirely different behaviors. The fourth is consistency. What a leader chooses in the single moment when schedule deadlines and safety collide is remembered more powerfully than years of safety campaigns.
Decisive here is psychological safety. As research by Harvard Business School's Amy Edmondson shows, without the assurance that reporting problems will not lead to disadvantages, an organization loses risk information. An organization where risk information fails to surface easily deludes itself into thinking it is safe until an accident occurs. The concept of a 'Just Culture' presented by James Reason similarly targets this point: the distinction must be established that honest mistakes are treated as learning materials while intentional rule violations are handled strictly, allowing reporting and accountability to coexist.
Safety Culture is Not Declared, It Grows in Stages
There are two representative frameworks for diagnosing safety culture. One is Patrick Hudson's maturity model, which categorizes organizations into five stages: Pathological, Reactive, Calculative, Proactive, and Generative. The pathological stage is characterized by an attitude of "as long as you don't get caught," the reactive stage by a state of moving only after an accident happens, and the calculative stage by having systems and documents in place but remaining at a formalistic level. Only at the proactive stage do organizations begin to proactively seek out risks, and at the generative stage, safety is embedded into the very method of work execution.
The other is DuPont's Bradley curve, which presents four stages: Reactive, Dependent, Independent, and Interdependent. The dependent stage maintains safety through rules and supervision, the independent stage is where individuals look after safety themselves, and the interdependent stage involves mutual intervention even in the safety of colleagues.
What both models state in common is clear: culture cannot be skipped. Attempts to transplant psychological safety into an organization lacking procedures and documents will fail, and conversely, it is a common trap for organizations lingering at the 'calculative stage'—having completed only documentation—to mistake themselves for being advanced. It can be diagnosed that a significant portion of domestic companies reached this calculative stage during their response to the Severe Accidents Punishment Act. The fact that the lack of procedures for identifying and improving hazard factors was most frequently cited in judgments suggests that not a few companies have failed to even fully reach that stage.
The 2026 Regulations Already Demand 'Culture' by Law
Notably, the amended Occupational Safety and Health Act of 2026 has converted core elements of culture into legal obligations. The revised risk assessment regulations, which took effect on June 1, 2026, stipulate that business owners must involve workers through methods such as round-table inspections, surveys, and interviews, and must also involve worker representatives if requested by them. Furthermore, they mandate notifying workers of assessment schedules, identified hazard and risk factors, improvement measures, and their implementation results through safety and health training, briefings, workplace postings, or written or electronic methods. It prescribes endeavors to constantly communicate factors that could lead to severe accidents through Tool Box Meetings (TBM) and the like before work.
Sanctions have also been newly established. Failure to conduct a risk assessment incurs a fine of up to 10 million won, failing to involve workers or worker representatives or failing to notify them of results incurs a fine of up to 5 million won, and failing to record and preserve results incurs a fine of up to 3 million won. However, under the addenda, fines will apply starting January 1, 2027, for workplaces with 50 or more regular employees (5 billion won or more in construction project value), and starting January 1, 2028, for those below that threshold. This gap between the implementation date and the sanction date serves as a practical preparation period.
Along with this, the scope of accident cause investigations has been expanded, provisions making accident investigation reports public after indictment have been introduced, and a system for the participation of honorary industrial safety supervisors during regional employment and labor office workplace inspections applied starting August 1. Subsequent legislative tasks also remain, such as the composition of joint primary-subcontractor occupational safety and health committees, the relaxation of requirements for exercising the right to stop work, and the establishment of penalty surcharges for corporations experiencing three or more fatal accidents annually, as previewed by the "Comprehensive Labor Safety Measures" announced on September 15, 2025. In short, the system has begun translating the two pillars of safety culture—'worker participation' and 'information disclosure'—into legal language and imposing them.
What to Measure: From Lagging Indicators to Leading Indicators
The practical bottleneck in building a safety culture is measurement. The indicators managed by most companies are lagging indicators, such as accident rates, fatality rates per 10,000 workers, and frequency rates. Because these aggregate already occurred results, they fail to indicate directions for improvement and have been consistently criticized for potentially acting as incentives for concealment.
Conversely, leading indicators capture the state of the organization prior to an accident. These include the number of near-miss reports and their processing turnaround times, the percentage of workers participating in risk assessments, completion rates and average delay days for improvement measures, the number of work stoppage requests, the frequency of management site visits, and the volume of hazard information shared with partner companies. An increase in near-miss reporting should be read not as a sign of deterioration, but as a sign that reporting channels are alive, and this shift in interpretation is itself a matter of leadership.
The Occupational Safety and Health Disclosure System can act as a lever at this point. The disclosure items—'previous year's activity performance,' 'current year's activity plan,' 'safety and health investment status,' and 'recurrence prevention measures and implementation plans'—inherently possess the character of leading indicators. If companies must establish systems to constantly accumulate such data precisely to fulfill disclosures, a structure is created where the system demands the infrastructure for culture formation.
Global Standards Are Also Moving in the Same Direction
International standards are likewise moving within the trend of expanding safety from systems to culture and well-being. ISO 45001, enacted in 2018, embarked on its first revision process in mid-2024, with a Draft International Standard (DIS) released in mid-June 2026 and member country voting and feedback collection underway. According to certification bodies, this voting closes on September 8, 2026, and final publication is projected for 2027. A three-year transition period is expected to apply. New definitions such as 'work-related well-being,' 'disability,' and 'external provider' are known to have been added at the DIS stage, demonstrating that the revision direction is extending beyond physical safety to well-being and supply chains. However, since final wording may change at the FDIS stage, it is premature to conclude these as finalized contents.
In the area of sustainability disclosure, industrial safety is a core social (S) metric. GRI 403 provides disclosure standards related to occupational safety and health, while ESRS S1 under the European CSRD framework requires the disclosure of health and safety indicators for own workforce. In the case of the ISSB, it added a human capital research project to its work plan in April 2024 and received research results on the necessity and feasibility of standard-setting at its December 2025 meeting, though a decision on whether to establish standards has not yet been made. The data accumulated by the domestic occupational safety and health disclosure system has a high potential to connect with these international disclosure demands going forward.
Six Steps Businesses Must Initiate Now
① Confirm disclosure target status first. The criteria of 500 or more regular employees or an annual construction project value of 120 billion won or more must be judged at the legal business owner level rather than the group level, and data collection systems must be designed now by counting backward from the first regular disclosure deadline of April 30, 2027.
② Reorganize risk assessment procedures based on revised laws. Worker participation methods, response procedures upon worker representative requests, methods for communicating results, and their documentation must be established. The crux is whether it was practically communicated to shift workers and foreign workers, and mere attachment of notices cannot be regarded as fulfilling the obligation.
③ Build an integrated data system including subcontractors and partner companies. The fact that fatal accidents of affiliated supplier workers are a disclosure item under the Enforcement Decree means that the safety performance of subcontractors is directly reflected in the primary contractor's external report card.
④ Operationalize evaluation criteria for occupational safety and health managers. This is the deficiency most frequently cited in judgments; records of actual evaluations and management conducted at least once every half-year are required, rather than formal evaluation sheets.
⑤ Form a leading-edge indicator dashboard. Simply placing near-miss reports, improvement completion rates, and work stoppage requests on monthly executive meeting agendas changes the organization's focus.
⑥ Regularize safety culture diagnoses. Measure psychological safety and reporting culture levels through anonymous surveys, apply the results to maturity models to check the current stage, and set tasks for transitioning to the next stage.
KBR Insight
The fact that accident fatalities in the first half of 2026 hit a record low since statistics began is a clear advance. However, upon dissecting the contents, optimism is premature. Fall accidents, where the effects of physical controls appear relatively directly, decreased significantly, but fires and explosions involving entangled multiple management factors actually increased. It can be seen that a divergence is occurring between areas where external pressures like supervision and inspections yield relatively swift effects and areas that improve only when internal management capabilities and culture within organizations change.
Judgments over four and a half years of the Severe Accidents Punishment Act tell the same story. What was repeatedly pointed out in court was not the absence of sophisticated technology, but the absence of procedures to find and fix hazards, and criteria for evaluating responsible personnel. KBR assesses that this is an issue difficult to resolve solely through criminal sanctions, and one that can only be solved when management's time and budget allocations move in tandem.
This is why we focus on the Occupational Safety and Health Disclosure System. We believe the core of this system lies not in intensified punishment, but in the creation of comparability. The moment safety investment amounts and subcontractor fatality statuses of companies in the same industry are placed side by side, safety goes beyond a simple cost item to increasingly influence reputation and capital-raising conditions. However, the risk that the system may end up in a competition of formalistic documents also persists. The capacity to fill out disclosure forms and the actual safety of the field are separate matters, and bridging this gap is precisely leadership.
Ultimately, safety culture is not built by campaigns or slogans. What management chooses at the moment schedule deadlines and safety collide, and what happens to the person who reports a problem—those two things determine an organization's safety culture. The system has now begun demanding that the results of those choices be disclosed annually. In April 2027, when the first report cards are revealed, the market will place companies' answer sheets side by side and compare them.