Jean-Jacques Pluchart The digital transformation of the industry involves mobilizing innovative artificial intelligence techniques. ​These techniques use the Internet of Things (IoT), cloud computing, data exchange, prescriptive analysis, new business models, etc. They apply advanced methods to manage data flows from heterogeneous systems.  ​They aimto achieve greater ​energy efficiency, more efficient maintenance, protection against breakdowns or intrusions, etc. In principle, they enable better safety, productivity, quality and profitability of industrial systems. ​They operate on three levels: that of capturing operational data from suppliers and customers; that of connecting stakeholders; that of transforming data into decision-making aids and valuable actions. AI offers continuous analysis capabilities dedicated to the collection of sensory data, fault diagnosis, flow modelling and the prescription of valid solutions. ​Current research focuses in particular on the integration of critical maintenance, safety and cybersecurity processes. ​They strive to improve the performance of systems without compromising their security. ​Systems engineers must choose the ​most suitable learning, optimization or prediction methods for the machines’ fields of application. ​This is particularly the case in the electrical energy sector. ​  ​ ​ The digital management of the processes of generation, transport, distribution and consumption of energy resources helps to reduce the mechanical inertia of the electricity network and to better ensure the balance of power between production and consumption. ​AI makes it possible to capture, store and process an increasingly large mass of data in order to make “the network smarter”. ​In the nuclear industry, AI makes it possible to improve predictive maintenance (by means of vibration sensors, real-time alerts), anti-collision detection and monitoring of sensitive sites. Among the digital techniques implemented in all industrial sectors, that of digital twins is emerging as a major lever for operational optimization. ​The digital twin is an interconnected system, powered by data from IoT systems, supervision platforms and simulation software. ​By building a virtual model of real objects, this technique offers companies increased visibility into their processes, better predictive maintenance and faster development of new products, without impacting production. ​However, it creates cybersecurity problems, as it reveals the “trade secrets” and “industrial comparative advantages” of innovative companies. ​It exposes them to espionage, sabotage, manipulation of optimization parameters and/or destruction of critical data. ​The complexity of digital twins makes them difficult to secure, as they combine  ​heterogeneous software from a variety of vendors, integrating different IoT sensors, AI layers, physical simulators, edge tools and, above all, cloud computing. In the current context of software between advanced industrial states, these actions constitute major threats to their strategic resources. ​Thus, the digitization of industrial processes raises questions of national sovereignty that invite public and private decision-makers to extend the European directives on IT security, and in particular, and to adapt the personal data protection regulation (GDPR) to the industrial environment 4.0. In 2016, the Turgot club chronicled one of the first works devoted to the birth of “Industry 4.0 “. Kohler D., Weisz J-D. (2016), Ambition industrie 4.0. ​The challenges of the digital transformation of the German industrial model, Eds Eyrolles. Since the 1990s, German industry has been engaged in a “cobotics” or collaborative robotics approach combining robotics, mechanics, electronics and cognitive sciences to assist the operator of a machine. ​Since the 2000s, it has also initiated a process of “globotics” or globalization of resources thanks to AI. ​The latter makes it possible to shorten value creation chains and decision-making circuits within organizations and their ecosystems, but it also accelerates the phenomenon of job relocation in laboratories, offshore factories or call centres. ​It also promotes the emergence of new forms of open organizational innovation based on free software, co-working and distance working, in principle more agile and less expensive, which extend from research and development (living labs, fablabs, etc.) to cooperative production (digital micro-manufacturing, do-it-yourself, maker spaces, etc.), and collaborative consumption (peer-to-peer accommodation, car sharing, etc.).

From symbolic AI to connectionist AI Jean – Jacques Pluchart The history of AI is marked by a tension between two approaches, alternately symbolic and connectionist, as observed by Cardon, Cointet and Mazières (2018). The researchers, following Lecun (2015), relaunched AI by processing massive data using so-called “deep neural” models (deep learning) and following a logic borrowed from cybernetics. This approach, described as “generative”, “inductive” or “connectionist”, has long been marginalised after the launch of symbolic AI in 1956 at Dartmouth by John McCarthy and Marvin Minsky, followed by the development of expert systems before the emergence of machine learning in the 1980s. Symbolic models were developed by a limited number of heavy league researchers, composed of a group of MIT (Minsky, Papert), Carnegie Mellon (Simon, Newell) and Stanford University (McCarthy), which mainly responded to public tenders and engaged in more or less playful experiments: chess or go games, dynamic simplified spaces, simulation of sets, semantic networks, truth functions, robotisation of behaviours, creation of new languages, etc. While symbolic AI applies a model to data following a hypothetical-deductive reasoning, connectionist AI follows an inductive logic by applying a learning method that makes it possible to make predictions by iteration of massive data. While symbolic AI applies a model (a theory or a heuristic) to structured data in order to verify a result at a given horizon, connectionist AI produces original content by “learning data” through appropriate questioning. While symbolic AI attempts to solve a predefined problem, connectionist AI induces meaningful representations from the interactions between social actors. This approach follows the logic of cybernetics initiated in 1948 by Norbert Wiener. The renaissance of connectionist AI is attributed in particular to the Parallel Distributed Processing research group led by Rumelhart et al. (1986). The work of the PDP explores the deep mechanisms of knowledge by exploiting the metaphor of neurons (a network of connections) and assuming that it is constructed by a binary activation mechanism. For more than 60 years, this controversy between researchers on AI has given rise to countless scientific works since, according to Cardon, Cointet and Mazières (2018), the “symbolic” corpus totalled 65,522 publications between 1956 and 2018, while the “connectionist” corpus gathered 106,278 publications. This vast debate is part of a process of scientific construction and deconstruction theorised in particular by Latour (1988). Références CARDON D, COINTET J-P. et Mazières A. (2018), « La revanche des neurones. L’invention des machines inductives et la controverse de l’intelligence artificielle », Réseaux 2018/5 (n° 211). LATOUR  B. (1988) , Science in Action: How to Follow Scientists and Engineers Through Society , Harvard University Press. LECUN Y., BENGIO Y., HINTON G. (2015), « Deep learning », Nature, vol. 521, n° 7553. RUMELHART D. E., McCLELLAND J. L. (1986), « PDP Models and General Issues in Cognitive Science », in PDP RESEARCH GROUP (1986), Parallel Distributed Processing. Explorations in the Microstructure of Cognition, Cambridge MA, MIT Press. WIENER N. (2014), La cybernétique : Information et régulation dans le vivant et la machine, Seuil

The law firm Debouzy organised a series of conferences on the challenges of generative AI, aimed at corporate lawyers and managers. On 18 November 2025, the Turgot club was invited to the conference led by lawyers Desrousseaux and Pérot, specialists in patent and copyright law. The two lawyers argue that the legal problems differ according to the phases of the process of value creation by AI: the collection of data by AI agents, the learning by software (in the form of texts, images, voices or sounds, videos, codes) and the exploitation of applications thanks to user questions (prompts). The EU General Data Protection Regulation (GDPR), enacted in 2016, governs how the personal data of natural persons can be processed and transferred in Europe. It was reinforced in 2024 by certain provisions of the European AI Act. In the United States, data is protected in particular by the Uniting and Strengthening America by Providing Appropriate Tools Required to Intercept and Obstruct Terrorism Act (USA Patriot Act). In principle, the data is under opt-out and is not subject to opt-in (protected data that cannot be used). They can then in principle be used in the learning phase of the application, but in the operating phase, they cannot be reproduced in full in response to user prompts. They can only be partially transformed or reproduced. Digital markings of creations or inventions make it possible to better identify the data processed. At the stage of exploitation of the results of the application, it is difficult to measure their degree of creation or invention. The two lawyers note an increase in litigation for infringement as well as negotiations or transactions for the sharing of the value created by AI from patented inventions or protected intellectual creations. The trend would be towards negotiation rather than litigation because infringement is difficult to prove. Overall, lawyers believe that current patent, copyright and professional secrecy regulations are sufficient to protect inventors and creators. The increase in the number of cases decided and cases of use should be sufficient to establish jurisprudence and stabilise practices. Jean-Jacques Pluchart

On 25 November 2025, the Turgot club, represented by J-J.Pluchart, was invited by the Cerfrance network to lead a conference-debate on the current challenges of AI for productive companies and professional associations. The Cerfrance organisation brings together 700 agencies, 14,000 experts and 320,000 clients “committed to economic, human and sustainable performance”. After having retraced the cycles of development of AI, successively “weak” then “strong”, the functionalities and the trades generated by AI were analysed. The AI “technology stack” now comprises three layers: systemic AI (semiconductors, networks, data centres), functional AI (recognition, data management, processing, storage, security), and agentic AI (Text: translations, article writing, summaries, scripts, automatic responses; Image: artistic creation, graphic design, illustrations; Video and audio: generation of synthetic voices, music, animations; Code: programming assistance, generation of scripts or functions). More and more sectors of activity must adapt their work organisation, in particular health, defence, banking and insurance, cybersecurity, industrial production (cobotics), mobility, publishing, cinema and media, education, research, programming, etc. Strategic functions, crisis management and local services are a priori spared. The adaptation proposals formulated in the reports of Villani (2018), Draghi and Letta (2023), Aghion-Bouverot (2024) and the AI Summit of February 2025 (AI “power lever”) were presented. The issues raised by the development of AI were then discussed: AI and national sovereignty, the contribution of AI to productivity and economic growth, the impact of data centres on GHG emissions, the financing of investments in AI R&D, the profitability of general AI models, the AI “stock market bubble”, the destruction and transformation of jobs, the dominant positions and cooperation agreements of GAFAM, the ethical codes of AI… It appears that the legal problems differ according to the phases of the process of value creation by AI: the collection of data by AI agents, learning by reinforcement models and the exploitation of applications thanks to user questions (prompts). The conference also focused on the general and specific biases of AI models: technical and psychological biases (perceptual, emotional and cognitive), specific biases of generative AI (related to linguistic disparities and implicit character), voluntary biases (simulations, manipulations, falsifications, intrusions)*.  The presentation was concluded with a reflection on the strategies to be implemented in order to transform AI into a competitive advantage, and in particular, on the new approaches to ‘augmented management’, the integration of management systems, the audit of ‘high AI quotient’ functions, new decision-making aids, benchmarking actions, nudging, assistance with quotes (Relief) and troubleshooting, quality control, cybersecurity, and training in AI and ethics. The conference was illustrated by use cases and references to the works and reports chronicled on clubturgot.com and analysed in the last book of the Turgot cub: New reflections on the wealth of nations. The lessons of Turgot and Smith. Jean-Jacques Pluchart *see review on AI and intellectual property

When one rereads the economic history of France, it seems that in every era, a voice has reminded us of the same simple truths: one must not spend what one does not have, one must not play with money, and one cannot build prosperity on promises or wagers. Turgot, Jacques Rueff, and Jean-Marc Daniel embodied this fidelity to reality. Each, in his own century, defended the idea that balancing public finances is not a constraint but a condition of freedom. Their legacy has endured through time, tracing a red thread from the eighteenth century to the present day. Turgot: Rigor as a Starting Point When Turgot was appointed Controller-General of Finances, he found a state in fiscal disorder. He rejected privileges and denounced waste. His guiding principle can be found in his Reflections on the Formation and Distribution of Wealth (1766), where he wrote: “We do not create wealth by distributing what we do not have.” This sentence, often quoted, expresses more a philosophy than an economic theory: rigor is not an accounting obsession, it is a moral requirement. For him, a deficit is an injustice passed on to future generations. In a France bogged down by rents and privileges, he defended the freedom to work, the free circulation of grain, and the abolition of forced labor. Jacques Rueff: The Guardian of Monetary Truth A century and a half later, Jacques Rueff took up the torch. He too lived in a world of illusions — those of a misinterpreted Keynesianism and of deficit financing through monetary creation. Alongside General de Gaulle, he helped lead the 1958 fiscal recovery of the French economy. For him, public debt was not just a number but a political fault. In The Social Order, he wrote: “No order can be built by defying the natural laws of the economy.” Budgetary balance, in his eyes, was an instrument of sovereignty: every deficit, every indulgence, led to dependency. He saw money as a moral instrument before being a financial one. In The Relentless Problem of Balance of Payments, he extended Turgot’s spirit: without fiscal discipline, there can be no lasting freedom. Rueff rejected fatalism. In Unemployment and Money, he demonstrated that unemployment results from accumulated rigidities. He advocated greater labor-market flexibility and the defense of free competition. Jean-Marc Daniel: Growth Through Freedom and Responsibility Jean-Marc Daniel stands in this same lineage. A liberal economist, he rebukes Keynesian complacency. His work belongs to a globalized, open economy where the temptation of protectionism and public spending remains strong. His intellectual mission: to remind us that sustainable growth rests on four essential pillars — work, saving, freedom (competition), and education. For him, rigor goes hand in hand with pedagogy. The State cannot produce wealth; it can only guarantee the conditions for it: security, justice, education, and monetary stability. Economics, in his eyes, is not a machine — it is a moral order founded on the truth of prices established by free competition and the reward of effort. In this sense, Daniel is a continuator of Turgot and Rueff: he denounces public budgetary excesses and insists that prosperity cannot be decreed — it must be learned. Philippe Aghion: Creative Continuity Philippe Aghion extends this heritage into another dimension — that of innovation. Where Turgot saw rigor as the condition of freedom, and Rueff viewed sound money as the keystone of prosperity, Aghion introduces the discipline of creativity. Inspired by Schumpeter, he formalized creative destruction: progress does not come from a spendthrift state but from a stable framework in which firms are free to innovate, fail, and begin again. Like his predecessors, Aghion does not oppose innovation and rigor — he connects them. Without strong institutions, quality education, and incentives for effort and investment, there can be no lasting progress. In this sense, he continues the spirit of Turgot and Rueff: freeing human energy while maintaining the discipline of rules. He also joins Daniel in emphasizing the importance of knowledge and education. Conclusion From Turgot to Rueff, from Daniel to Aghion, four voices, four centuries, one lesson: discipline — whether fiscal, monetary, or intellectual — is not optional; it is a political necessity. Economic disorder always prepares social disorder and, ultimately, leads to servitude, while rigor opens the path to freedom. Turgot’s thought has endured because it is rooted in reality and rejects illusions. Its legacy endures because success does not lie in mortgaging the future but in giving it every chance to exist free of servitude. To reread Turgot, Rueff, Daniel, and Aghion is not to yield to nostalgia; it is to rediscover, beneath today’s debates, the keys to lasting prosperity. Benoit Frayer November 2025 References Anne-Robert Jacques Turgot, Reflections on the Formation and Distribution of Wealth (1766). Jacques Rueff, The Social Order (1945); The Relentless Problem of Balance of Payments (1965); Unemployment and Money (1931). Jean-Marc Daniel, Capitalism and Its Enemies (2016); The Collusive State (2014); A Living History of Economic Thought (2018). Philippe Aghion, The Power of Creative Destruction (with Céline Antonin and Simon Bunel, 2020); Endogenous Growth Theory (with Peter Howitt, 2008). Joseph Schumpeter, Capitalism, Socialism and Democracy (1942) — theoretical foundation of “creative destruction,” later extended by Aghion. Benoit Frayer November 2025

Jean-Jacques Pluchart The digital economy, and in particular Artificial Intelligence, are often presented as freeing the worker from the most alienating tasks in favor of more creative actions, but they are also perceived as being able to generate a loss of meaning of action, professional malaise and violence at work. In the context of an organization, violence can take many forms: verbal and physical, psychological and social, symbolic and structural, which differ according to multiple factors: the activity carried out, the work situation, gender, but also according to the systems implemented, as in the case of digital technologies covering automation and expert systems, the Internet and social networks, symbolic and generative AI applications.  Violence can be exercised between the actors themselves (between colleagues, between superiors and subordinates) and/or between the latter and the stakeholders (customers, users, suppliers, etc.) of the company or the administration, but it can also be caused by a procedure or a system. The most frequently cited form of violence against workers generated by AI is the fear of losing one’s job and thus being socially downgraded, or of having to adapt to a new job that is said to be “augmented” by AI.  The future “robot-man” fears, in particular, being confronted with the ingratitude and loneliness of a job carried out remotely, alone in front of a screen, prey to the dysfunctions and “black boxes” of a system, and most often subjected to digital panoptism. They fear losing the meaning of their work, no longer recognizing their symbolic order, no longer knowing their professional identity. He fears being exposed to the stress and burnout of the ‘enslaved man’. This anxiety can be all the more depressive as he can no longer activate his defense systems (by denial, displacement, derision, sublimation…) against a “robot” whose grip is inevitable.    The violence of this new relationship to work is all the more implicit as it is marked by the uncertainty weighing on the date and conditions of the implementation of the new system thus perceived as a ‘black swan’. The threat is all the more latent as it covers a growing number of jobs, ranging from back office (administration) to middle office (production and control) and front office (customer relations, etc.). It now reaches managers and executives responsible for reorganizing a company or a service, arbitrating between often complex operating systems, ensuring their cyber security and training staff in new practices. They are thus exposed to new types of risks to the sustainability of their organizations and to the future of their own careers. These incomplete observations show that the forms of violence at work generated by the accelerated development of AI can only be detected, analyzed and framed by HRM approaches using psychology and sociology, but also anthropology and psychoanalysis.        

Anne Robert Jacques Turgot Chronique de François MeunierTurgot had said almost everything as early as 1766, ten years before Adam Smith, about the formation of the interest rate and its relationship with the return on capital and the price of fixed-income financial securities. This is in his major economic work: “Reflections on the Formation and Distribution of Wealth .” In particular, three concepts that are not simple to understand or to make understood. 1. That the rate of return or cost of capital (depending on whether we take the point of view of the investor or the company) depends on the associated risk. The interest rate (which ensures a fixed income, independent of the profitability of the investment except in the event of default) must then be lower. (Underlining and brackets are from the author)I have counted five different ways to use capital or to invest it in a profitable way. The first is to buy land that yields a certain income. The second is to invest one’s money in agricultural enterprises by leasing land whose fruits must yield, in addition to the price of the rent, the interest on the advances and the price of the labor of the person who devotes his wealth and his effort to cultivating them. The third is to invest one’s capital in industrial and manufacturing enterprises. The fourth is to invest it in commercial enterprises. And the fifth, to lend it to those who need it, for an interest. […] It is obvious that the annual products that can be withdrawn from the capital invested in these different jobs are limited by each other, and all relative to the current rate of interest on money. • LXXXIV. — Money invested in land must yield less. Anyone who invests their money by buying land leased to a well-paying farmer obtains an income that gives them very little trouble to receive, and that they can spend in the most pleasant way by giving a career to all their tastes. It also has the advantage that land is the most secure possession against all kinds of accidents .• LXXXV. — Money lent must yield a little more than the income of land acquired with equal capital. He who lends his money at interest enjoys even more peacefully and more freely than the landowner; but the insolvency of his debtor can cause him to lose his capital. He will therefore not be satisfied with an interest equal to the income of the land he would buy with the same capital. The interest of the money lent must therefore be stronger than the income of land purchased for the same capital, because if the lender were to buy land of equal income, he would prefer this use.• LXXXVI. — Money invested in agricultural, manufacturing, and commercial enterprises must yield more than the interest on money lent. For a similar reason, money employed in industry or commerce must yield a more considerable profit than the income of the same capital employed in land or the interest of the same money lent; for these employments require, in addition to the capital advanced, a great deal of care and work, and if they were not lucrative, it would be much better to obtain an equal income that could be enjoyed without doing anything. […] 2. That the costs of capital are interrelated, in a proportion that depends on the risk. An increase in the cost of capital causes an increase in the interest rate, through an arbitrage relationship. In modern terms, this resembles the CAPM. • LXXXVII. — However, the products of these different uses are limited by each other and are maintained despite their inequality in a kind of equilibrium. The different uses of capital therefore yield very unequal products; but this inequality does not prevent them from reciprocally influencing each other, and from establishing a kind of balance between them. […] I suppose that suddenly a very large number of landowners want to sell their land: it is obvious that the price of land will fall, and that with a smaller sum one will acquire a greater income. This cannot happen without the interest of money becoming higher; for the owners of money will prefer to buy land than to lend it at an interest that would not be stronger than the income of the land they would buy. So if borrowers want to have money, they will be forced to pay a higher rent. If the interest of money becomes higher, it will be better to lend it than to assert it, in a more painful and risky way, in the enterprises of culture, industry and commerce, and we will only do business with those who will bring, in addition to the wages of labor, a much greater profit than the rate of money lent.In short, as soon as the profits resulting from any employment increase or decrease, capital is poured into it by withdrawing from other employments, or is withdrawn from it by pouring into other employments; which necessarily changes the ratio of capital to annual product in each of these employments. […] The product of money used in any way whatsoever cannot increase or decrease without all other uses experiencing a proportionate increase or decrease. 3. Finally, the price of annuities or fixed-income bonds is inversely related to the interest rate. LXXXVIII. — The current interest rate of money is the thermometer of the abundance or scarcity of capital; it measures the extent that a nation can give to its cultural, manufacturing and commercial enterprises. […] It is obvious that the lower the interest on money, the more valuable the land. A man who has fifty thousand pounds of annuities, if the land is only sold at twenty [i.e., with a P/E of 20X or a rate of return of 5%], has only a wealth of one million; he has two million if the land is sold at forty [a P/E of 40X]. If the

On September 12, Fitch downgraded France’s sovereign debt from AA- (high quality) to A+ (upper average quality). In its analysis, the agency uses GDP as a central indicator to assess the sustainability of public finances, in particular through the debt/GDP ratio. Fitch points out that French public debt is expected to rise from 113.2% of GDP in 2024 to 121% in 2027, with no clear prospect of stabilization. Fitch considers it unlikely that France will bring its public deficit below 3% of GDP by 2029, while the government was aiming for this objective to comply with European rules. The deficit is expected to be 5.4% of GDP in 2025, and should remain above 5% in 2026 and 2027. In this context, the agency believes that France has less room for maneuver in the face of possible economic shocks. According to the latest INSEE estimates, the country’s GDP will grow by 0.8% in 2025 instead of the 0.6% initially forecast, due to better dynamics in agriculture, tourism, the real estate market, and aeronautics. This increase, although positive, is far from expected to offset the weight of the debt. In its latest economic bulletin published on September 25, the ECB presents an overview of the macroeconomic projections of inflation and GDP in the Eurozone, in order to justify its monetary policy. Inflation is expected to be 2.1% in 2025, 1.7% in 2026, and 1.9% in 2027. Growth, meanwhile, is expected to be 1.2% in 2025, 1% in 2026, and 1.3% in 2027. In its statement, “the Governing Council considers it essential to strengthen, without delay, the eurozone and its economy in the current geopolitical environment. Fiscal and structural policies should improve the productivity, competitiveness and resilience of the economy… It is up to governments to prioritize structural reforms and strategic investments that promote growth while ensuring the sustainability of public finances.”The roadmap is therefore clear: GDP must be supported However, is the calculation of GDP still relevant and up-to-date to measure the wealth created by a country? Should its formula evolve as suggested by some economists and politicians? And especially in terms of climate change? GDP is already 80 years old It was at the Bretton Woods conference in 1944 that GDP was adopted as a standard indicator to measure the economic activity of countries, in particular to facilitate international comparisons and post-war reconstruction. It will thus be gradually adopted by the whole world and international organizations, such as the UN, the IMF or the World Bank. In France, it was applied from 1949. Over the years, GDP gradually replaced GNP (Gross National Product) as the main indicator, because it measures production in a territory, regardless of the nationality of the economic actors. In the 70s and 80s, GDP began to include previously neglected sectors such as services. Indeed, developed countries are moving from an industrial economy to a service economy, which changes the structure of GDP The limits of GDP In 2009, the Stiglitz-Sen-Fitoussi report or the “Report on the measurement of economic performance and social progress” was commissioned by Nicolas Sarkozy. Questioning the limits of the calculation of the indicator, this report proposes to supplement the GDP with indicators of well-being, sustainability and inequalities. Indeed, the report points out that GDP: – ignores well-being: it does not measure happiness, health, education, or the quality of the environment;– neglects inequalities: an increase in GDP can hide an increase in income gaps. It ignores sustainability: It does not take into account the degradation of natural capital (resources, climate);– does not value unpaid work, such as volunteering or domestic work.Recently, the ECB also warned its member states that climate change could cut European GDP by 5% by 2030. In France, INSEE proposed in 2024 a complementary indicator to GDP, the “adjusted net domestic product” (Pina): this indicator makes it possible to take climate change into account by netting the creation of value from the effects of future damage and decarbonization costs. Although developments are being studied and considered, GDP is still the reference for a large number of institutions as a management tool; but other dimensions of analysis will have to be added in order to have a more global vision in a context of global warming. Sophie Friot Member of the Turgot Club

Part 2: New Approaches The manipulations observed in the social accounts and the sustainability reports of the companies are generally explained by models derived from the “Cressey triangle” or by the psychological biases analyzed by Kahneman and Sversky. But new approaches borrowed from phenomenology and psychoanalysis now allow to explain the development of managerial practices in “gray areas”. Traditional approaches Fraud covers intentional behavior contrary to laws, regulations, and financial, social, and environmental standards. It has been the subject of much research since the founding work of Sutherland, author of the famous “white collar crime” formula. The reference model applied to its various practices is that of the “fraud triangle”, proposed by Cressey (1967), according to which the fraud process develops along three axes:  – “The opportunity” to commit an illegal act and/or contrary to the interests of an organization, which is often offered by privileged access to sensitive resources (data, systems, bank accounts …) insufficiently protected.  – The “motivation” of the fraudster, which covers different types of psychological bias and psychic affects: the need for money, the quest for recognition, ambition, the taste for risk, mimicry, addiction to fraud. – The “rationalization” of the fraudster’s behavior, which corresponds to the practices of adverse selection intended to mask the fraudulent acts and to thwart the confidence of third parties, in particular by accounting concealments, which reflect the “excuses” that the fraudster gives to himself: “he only borrows the money”; “He falsifies the accounts to save the company”. Cressey’s model has been adapted to new forms of management. Albecht (1982) distinguished three factors favorable to fraud: environmental pressure, opportunism, and the psychological profile of the fraudster. Rezaee (2002) designed the “3C model” (Choice, Conditions and Corporate Framework). Bealey (2000) observed the internal contingencies (the history of the company) and external (its institutional framework) of the fraud process. The Statement of Auditing Standards classified 25 different factors of corporate fraud risk, according to 3 axes: the personalities of the leaders, the economic environment of the company and its organization. The Cressey triangle was reinterpreted by Dominey et al. (2012), who propose a model no longer focusing on the fraudster but on his practices. Called the “fraudulent act triangle model”, it also has three facets: a more or less sophisticated methodology (a misappropriation of assets, a transfer of liabilities…), a concealment of fraud (a false accounting entry, a file destruction…), a conversion of the proceeds of fraud into exploitable assets (money laundering). According to Smith and Lewis (2011), corporate accounting manipulations are generated by managerial drifts based on four types of paradoxes: – organizing paradoxes, which occur when groups of actors oppose methods (accounting-real, fraudulent-non-fraudulent); – belonging paradoxes, which arise when a goal can be achieved by different means (accounting-real), or when there are conflicting goals (short or long term); – performing paradoxes, which arise from more or less conflicting interests between stakeholders; – learning paradoxes between tradition and innovation, which result in a “phygital” treatment (combining experience and algorithms) of accounting manipulations.    According to Boudon (1990), fraud or manipulation has become a social phenomenon marked by “mimicry effects” and “composition effects”, by which interactions between the types of actors (intentional and unintentional manipulators, fraudsters and non-fraudsters) lead to perverse effects contrary to the intentions of each.  According to Tversky and Kahneman (1974), the behaviors of corporate actors are subject to four classes of bias that have been reinforced by the development of Artificial Intelligence and that particularly affect communication to meet ESG principles. The first class covers cognitive biases that distort the processed data, their processing models, and the interpretation of the results, including familiarity and confirmation biases. Faced with an urgent decision or a complex problem, managers choose the option they think they can best control or the solution that mobilizes immediately available resources and/or involves easily controllable issues. They are also subject to biases of “conservatism” which reflect the tendency to overestimate information in line with their convictions (Festinger, 1957), or anchoring biases, which consist in discarding discordant or confusing information and to seek only those confirming their own choices (Goetzman and Pelès, 1997). The second class of heuristics transposable to AI deals with excesses of optimism and confidence. Managers tend to interpret the “solutions” provided by the applications as “self-fulfilling discourses” or “performative presentations”, which give them the illusion of controlling the situation. They are victims of overconfidence, usually accompanied by self-justification in the event of a bad decision. The decision-maker has the illusion that they “manage in compliance”, that they “master the ESG criteria”, that they “inspire the confidence of their stakeholders” … They believe they do not need advice; they rationalize past events a posteriori (retrospective bias); they attribute all the merits of a success (self-attribution bias), according to Roll (1986) … The third form of bias relates to the effects of imitation or conformity, which affect, according to Hong, Kubik and Stein (1994), designers influenced by socio-professional norms, or by the follow-up of pioneers, charismatic leaders or events. The fourth form of drift caused by generative AI covers perceptual and/or emotional biases, which can blur the mental representation of a phenomenon (Higgs, Dulewicz, 2002). Certain ambiguous or counter-intuitive solutions revealed by AI can induce different behaviors from one actor to another in the face of identical situations. These biases can distort individual decisions in business.  Loewenstein et al. (2001) have shown that the fear of an uncertain event is often motivated by the possibility – and not the probability – of its negative consequences; because the more “the latter are perceived as important, the more the affective prevails over the cognitive”. The respondents’ answers (presented in the section make it possible to distinguish three new approaches to gray areas within organizations, which have not yet been proposed – or which have only been mentioned – by researchers and experts on the issue of gray areas in management. This exploratory survey makes it possible to go beyond the traditional approaches, according to which (non-fraudulent) manipulations in

Jean-Jacques Pluchart The research workshop organized on June 27, 2025 by the Institute of Psychoanalysis and Management (IPM), an academic association member of the FNEGE, gave rise to several communications on the theme of “gray areas of the management of organizations”. Professor J-J. Pluchart (Scientific Director of the IPM) presented research on the gray areas of financial and extra-financial communication, the results of which are likely to interest the readers of clubturgot.com. Since its inception, financial accounting has given rise to various frauds contrary to the regulations and standards in force, to which have recently been added so-called “creative” practices apparently aligned with an accounting framework, but in fact not in accordance with the ethics of the company. These behaviors are part of the “gray areas” of management, located on the border between the regulatory and non-regulatory domains. These practices have diversified with the obligation to publish sustainability reports, which requires the reporting of several hundred extra-financial indicators, both accounting and statistical, as part of ESG (Environment, Social, Governance) reporting. The accounting gray area has thus extended to practices relating to green and social washing, nudging, faking, etc. These deviations have been facilitated by certain applications – so-called projective – of Artificial Intelligence. The research, based on a questionnaire survey of a population of statutory auditors and trusted third parties, identified common practices observed in the grey areas covering accounting and sustainability reporting, as well as analyzing the motivations of their authors.  The panel surveyed distinguishes between fraudulent manipulations contrary to regulations, which are a priori in “black areas”, and non-fraudulent manipulations contrary to professional ethics or company ethics, which constitute “gray areas”. The discriminating criterion that dominates the auditors’ responses is compliance or non-compliance with laws, regulations, and professional standards.  Accounting and Extra-Accounting Practices The respondents distinguish: – Accounting fraud (“black areas”), such as the recording of fictitious transactions (sales, purchases, movements of stocks, receipts or disbursements…) and the issuance of false invoices; the non-recognition or recording of actual transactions that do not comply with IFRS standards (such as the activation of advertising or training costs); the falsification of accounting documents (invoices, contracts, certificates, labels…), the deconsolidation of subsidiaries in debt and/or in deficit; the failure to publish the company’s accounts or the publication of only pro-forma accounts; the early recording of income or delayed expenses from one year to the next… –  Non-fraudulent accounting manipulations, such as the adjustment of discretionary accruals (optional allocations or reinstatements of depreciation and provisions) and/or the need for working capital; the application of the big bath technique during a change of management, by exceptional allocations of provisions that can be reinstated in the following years; the smoothing of results over several years in order to maintain a regular distribution of dividends, and/or to display results in line with forecasts; the change in the inventory valuation method in order to generate capital gains or losses; the activation of certain expenses (R & D, interest …) and their amortization over several years; the revaluation of certain assets (real estate, goodwill …) using models generating capital gains or losses (also cited by Chiapello, 2005); the unusual use of factoring or discounting to improve cash flow; the non-publication or partial publication of accounting results despite the risk of legal proceedings; the publication of “oriented” pro-forma results, in order to influence the course of the action; the manipulation of segmented information to guide comparisons between competitors in the same industry … – “Real accounting manipulations” (creative accounting), such as the artificial increase in sales through excessive year-end discounts and/or exceptionally favorable invoice payment terms; the deferral of expenses from one year to the next (including research and development and/or training expenses); the realization of lease back operations of various assets (headquarters, stores, warehouses, factories, equipment, etc.); the abnormal disposal of non-operating and/or investment assets… – Extra-financial manipulations, such as in black areas, non-compliant practices of disinformation, qualified as environmental (green washing) and social or societal (social washing) laundering, covering erroneous, imprecise or truncated data; in gray areas, non-information (some key data are omitted) or non-monitored information (companies display objectives but not results), and the biased framing of the company’s projects or operations:  – over time, with simulations and projections (facilitated by AI) to present the most favorable or most credible data (such as net-zero or very long-term gender parity objectives without regular monitoring of achievements);   – in space, with data (also processed using AI) not representative of a field, due to intentional targeting errors and/or biased parameters, ambiguities or textual inconsistencies, which lead to errors in data interpretation. Factors favorable to gray areas  Overall, the auditors interviewed believe that it is increasingly difficult to isolate the types of more or less fraudulent manipulations, insofar as a growing number of them (misappropriation of assets, fraud in purchases or overheads, etc.) are internal or external while benefiting from internal complicity, and are the subject of increasingly difficult to detect hedging manipulations. Manipulations that are deemed to be in gray areas by auditors and ICOs (Informative Commissionner Officers), are generally observed when: – the company is over-indebted, its results are declining, its stock price is volatile and/or the continuity of its operations is threatened;  – conflicts of interest between the company’s stakeholders (in particular between investors, partners, staff, the State, etc.); – the company’s shareholding is open and fragmented; the smoothing of the company’s results reassures its stakeholders about the company’s resilience; – the company’s image is less likely to be degraded if it complies with accounting rules but not or little with ESG standards, which are more recent and uncertain. According to the respondents, some managers therefore justify their intentions and behaviors by: – “good reasons”: the difficult or particular situation of the company justifies a “certain interpretation” of the rules and standards; – “professional routines”: “accountants have always practiced accounting and tax optimization, the new standard is inapplicable…” – “beliefs”: “non-information makes it possible to avoid green or social bashings; the superiority of the shareholder model, innovation