Beyond the Tools
When a newsroom decides to integrate artificial intelligence in a planned way, the transformation manifests both in team productivity and in internal conversations about who does what, who supervises what, and who has the authority to decide what gets delegated. In that process, a new morphology of journalistic work takes shape that goes far beyond the tools themselves.
For much of the twentieth century, newsrooms operated under relatively stable dynamics. Reporting teams investigated and wrote, editing teams reviewed and corrected, the deadline organized the day and publication completed the cycle at the end of the workday. Each function had its place in the chain and responsibilities were distributed around a conventionally accepted dynamic within media organizations.
The arrival of the Internet in the mid-1990s transformed rhythms, formats and the relationship with audiences. Real-time logic displaced the deadline as the great organizer of work, platforms multiplied distribution channels and newsrooms incorporated new profiles to operate in that environment. Even so, in that context of profound change, the productive structure of most media organizations remained intact — like an assembly line with differentiated roles and a division of labor with few major alterations.
Artificial intelligence is the continuation of that process, but with a different logic. It does not burst into that scheme to replace it abruptly — it reconfigures it through multiple entry points. An automatic transcription system changes the relationship with recorded material. A headline generation tool redistributes a task that was previously exclusively editorial. An agent that monitors sources in real time introduces a layer between teams and information that did not previously exist. None of these changes, taken separately, transforms a media organization. All of them together profoundly transform the internal structure that organizes them.
What emerges then is not an improved version of the traditional newsroom but a qualitatively distinct system. In that system, people with established roles and profiles that do not yet have names coexist, alongside manual and automated workflows, human decisions and processes that occur without direct intervention. The logic of "doing" shifts toward a logic of "orchestrating," which requires new skills to coordinate, supervise, validate and decide in environments where not everything that happens yet has a clearly assigned owner.
That coordination among heterogeneous parts is, in itself, a new competency. Newsrooms are learning to build it with variable resources, without consolidated models to imitate and in the middle of a transformation that does not wait to be understood.
Three movements, a new logic
The reconfiguration of work within media organizations does not operate through abrupt substitution but through progressive displacement. Three movements can be identified in this process that occur in an intertwined manner. The first is the automation of mechanical tasks: activities requiring repetitive execution or the application of predefined rules migrate toward automated systems. The second is the expansion of cognitive capabilities: tasks previously unviable due to limitations of time, scale or access to information become possible. The third is the redistribution of time and attention: the time freed by automation does not convert directly into more production, but migrates toward tasks of greater complexity, process supervision and decisions requiring irreplaceable human judgment.
These three movements do not occur in successive stages. They coexist, overlap and generate tensions. The same newsroom may be automating transcriptions while expanding its data analysis capacity and redistributing its reporters' time toward long-form investigations. Each of them has a concrete manifestation in daily work, and understanding them individually is the condition for managing them collectively.
The case of transcription illustrates well the nature of this first movement. For decades, transcribing interviews was a basic journalistic task that consumed a considerable amount of a newsroom's productive time. With high-precision automated transcription, that time was significantly reduced. The type of attention required changed, moving from mechanical execution to critical validation, but the human responsibility for final accuracy and the editorial judgment to identify what is relevant remains intact. The question that persists is whether something is lost in the process. One might argue that deep immersion in the material — the detection of subtle patterns that only emerge through slow and repeated listening — has no equivalent in review. There is no universal answer. What matters is that the task mutated, it did not disappear, and that mutation demands different competencies.
The second movement enables tasks that were previously unviable due to limitations of time, scale or access to information. The analysis of large volumes of documents for investigative journalism is the clearest example. Manually reviewing thousands of public contracts, leaked emails or court records could take weeks or simply be beyond the reach of a small team. With assisted analysis tools, processing time is considerably reduced and opens lines of investigation that previously depended on resources most newsrooms could not afford. What changes is not only the speed but the type of journalism that becomes possible — data coverage at local scale, systematic monitoring of institutional sources or the detection of patterns in public information that previously depended on conditions few newsrooms could sustain. While AI does not produce the investigation, it expands the territory a journalist can cover.
The third movement is the least visible and the most complex to manage. In newsrooms that have integrated AI consistently, it can be observed that total work time does not always decrease proportionally to the automation implemented. The time freed tends to migrate toward process supervision, coordination of hybrid workflows and error management when automation fails. The promise of doing more with less is only partially fulfilled — and in certain cases the opposite occurs. What seems to change is not so much the quantity of work but its nature, which requires less mechanical execution and more supervision, coordination and strategic decision-making.
This shift modifies the journalist's position within the production process. In the traditional newsroom, value was measured by what was produced. In the hybrid newsroom, a growing part resides in how one decides to produce: what gets delegated, what gets supervised and what is reserved for human judgment. Designing the architecture of that process requires solid editorial judgment, functional understanding of systems and the capacity to make decisions under conditions of uncertainty. It is, ultimately, a new way of doing journalism that demands, at the same time, a deeper understanding of its foundations.
The newsroom as a decision environment
The transformation this chapter describes also has an implication that organizations have not yet fully processed. The type of professional profile that newsrooms are beginning to need is changing.
For years, the digital evolution of journalism was associated with specialization. The growing complexity of formats, platforms and tools consolidated increasingly specific roles within the production chain. Reporters, editors, designers, developers and audience analysts each operated within a defined function in a system that, while digital, preserved an industrial logic based on division of labor.
Artificial intelligence puts pressure on that model. Not because it eliminates the need for specialization, but because it makes more valuable the profiles capable of connecting different parts of the process. Understanding how a story is produced, how it is distributed, how its impact is measured and where an automated system can intervene is becoming more relevant than the expert execution of a single task. That displacement generates friction within media organizations. Newsrooms today host profiles trained in the logic of specialization alongside others who join with different references and practices. Managing that coexistence becomes, in itself, part of the transformation.
Generalist profiles acquire new prominence in that context. It is not about professionals who know a little of everything, but about those who understand the system as a whole, can move between different stages of the workflow and act as translators between the editorial, the technological and the strategic. They are the ones who can decide what to delegate, how to integrate tools and where to intervene when the process breaks down.
This movement challenges the logic that organized newsrooms for decades. The idea of a production chain with fragmented tasks and stable roles loses centrality in the face of a more flexible dynamic, in which value lies in the ability to articulate, coordinate and make decisions about the complete process. The very notion of a "newsroom" begins to feel too small to describe what these organizations do. Designing, operating and supervising complex editorial systems requires a different work architecture. The newsroom transforms, in that process, into a decision environment.
That change does not occur only in workflows or in internal team conversations. It also occurs in org charts. Newsrooms that understood AI as a governance challenge made it visible by institutionalizing specific positions to manage it.
A new organizational layer
The morphological transformation of newsrooms is not a theoretical hypothesis. Between 2022 and 2026, at least 80 roles specifically linked to artificial intelligence were documented in media organizations across five continents, in a survey compiled for this guide. It is one of the most accelerated organizational transformations journalism has undergone in recent decades.
Behind these 80 cases there is something more than new job titles. An organizational layer is taking shape with its own logic, unprecedented hierarchies and profiles that did not exist three years ago. The structure of newsrooms is changing in terms of workflows and tools, but also in terms of who works within them and with what mandate.
The survey developed for this chapter was built from verifiable primary sources, including institutional announcements, published interviews, professional profiles and specialized media industry coverage. For a role to be included it had to have a formal designation, documented decision and attributable source. Those criteria also define the survey's limits. Informal experimentation, hybrid roles without an official name and transformations that occur without public announcement fall outside the list, suggesting that this process is probably more widespread than the documented cases reflect.
The new roles are grouped into four functional families. The first brings together strategic leadership profiles — senior positions with cross-cutting responsibility over the transformation. The second groups editorial production, where AI intervenes directly in the content workflow. The third concentrates positions related to infrastructure and product, connecting the newsroom with systems and readers. And the fourth brings together roles oriented toward training, governance and ethics within media organizations. None of these families existed as a consolidated category before 2020.
The four groups that organize the analysis are a reading tool, not a closed taxonomy. In practice, roles overlap. A Head of AI Strategy may design training programs and governance while simultaneously making decisions about infrastructure. What these broad groups allow is identifying the center of gravity of each position and observing which dimensions of the transformation organizations prioritized institutionalizing and which remain pending.
Each family is illustrated with four cases selected for geographic diversity, organizational scale and type of mission. The chart and the end of the chapter include an appendix with the complete list of the 80 documented roles, organized into categories.
Strategic leadership
This is the largest family in the survey and groups senior positions with cross-cutting functions over the transformation. They articulate editorial judgment with the technology agenda and report directly to the editorial or executive leadership of their respective organizations. Their numerical predominance is not coincidental — it indicates that the organizations that advanced furthest in AI integration understood it from the start as a governance challenge requiring supervision at the highest level, not as a technical project delegatable to a specialized team. The question these roles answer is not how to use AI but how to make decisions about it. In several cases that mandate extends to internal training and ethical governance — dimensions that, in the absence of dedicated roles, end up being absorbed by senior leadership.
The New York Times
Reuters
Helsingin SanomatEditorial production
This groups roles that redefine how journalistic content is produced when AI intervenes in the process. It includes both profiles that use AI as a central tool in their daily work and those who supervise systems that generate content at scale. The distinction between assistance and automation matters operationally, but in organizational terms both answer the same question: what does a journalist do when part of production is executed by a system. The most significant cases show that the answer is not uniform and depends on the type of content, the market and the editorial decisions each organization made about what to preserve for human judgment.
Folha de S.PauloNewsquest
News Corp AustraliaThe Connecticut MirrorInfrastructure and product
This groups engineers, product managers and technical profiles who work inside newsrooms with deep understanding of editorial workflows. What distinguishes them from traditional technology departments is the requirement to understand journalistic values in addition to technical architecture. They are the profiles who translate between what a system can do and what a newsroom needs, and make design decisions with direct consequences for the quality of the journalism the system produces or enables. In several of the most significant cases, this dual understanding was the result of extensive journalistic careers that later evolved toward the technological dimension of the work.
Associated Press
Nikkei
Scroll.in
Al JazeeraTraining, governance and ethics
This is the family with the fewest roles dedicated exclusively to this dimension, and that proportion is itself a relevant data point. In a context where the risks of editorial automation are growing, few organizations have created positions specifically oriented toward auditing, training and establishing responsible use frameworks.
It is worth noting a tension the survey makes evident. Several of the cases that best illustrate this dimension belong, in terms of primary mandate, to the strategic leadership family. They are included here because they document something equally significant: in the absence of pure governance and training roles, senior leadership positions are absorbing that mission. Ethics and training did not emerge as separate functions but as responsibilities that senior leadership incorporated by necessity and, in the best cases, by conviction. The cases that explicitly institutionalized this dimension show that the decision reflects an understanding of the transformation as cultural change rather than technical update.
AFPLenfest AI Collaborative
Bayerischer Rundfunk
APA AustriaWhat the survey shows as a whole
Beyond the classification by families, the survey allows identifying three patterns that cut across all of them and say something about the state of the transformation as a whole.
The first is speed. In February 2023, the Financial Times created what was probably the first formal AI Editor position at a major international outlet. Less than three years later, the survey documents 80 verified roles in more than 60 organizations across five continents. That pace of institutionalization has no recent precedent in the journalism industry — and it is occurring, moreover, in a context of widespread employment contraction in media.
The second is the concentration in leadership. That 30 of the 80 documented roles correspond to senior positions with cross-cutting scope is not a minor data point. It indicates that the organizations that advanced furthest understood AI integration as a governance problem before an operational one. The transformation was not delegated downward or to the margins of the structure — it was placed at the center of decision-making.
The third is the relative weakness of dedicated governance. Although the survey documents roles oriented toward auditing, training and establishing responsible use frameworks, their proportional weight within the total is significantly lower than that of the other families. In a context where decisions about what to automate have direct consequences for the quality and accountability of published journalism, that proportion signals an institutional gap that enthusiasm for adoption has not yet fully closed.
These three patterns are expressed concretely in the profiles the survey identifies as most influential, in the geographic distribution of the transformation and in the generational tensions that organizations are learning to manage. The following sections develop each of those dimensions.
The bridge profile
The survey's most relevant finding points to the type of profile that defines those with the greatest influence. In almost all cases of major organizational impact, the person in charge combines a journalistic career with functional understanding of technology. The NYT's Editorial Director of AI Initiatives arrived in the role after leading Quartz as CEO. Reuters' Newsroom AI Editor has 30 years of experience at the agency and today supervises prompt engineering. Folha de S.Paulo's Editora de Inteligência Artificial had 16 years of journalistic career before creating that section.
These are scarce, high-demand profiles with influence over the outcome of the transformation. Their scarcity is the result of decades of journalistic training that rarely contemplated this intersection as a possible professional destination.
The unequal distribution of transformation
The geographic distribution of the survey exposes asymmetries that are not accidental. The United States and the United Kingdom concentrate approximately 50% of the documented roles. Nordic countries have a striking presence given the scale of their markets, with ten positions across Sweden, Norway, Denmark and Finland. That overrepresentation reflects years of investment in institutional infrastructure for media innovation, including the Nordic AI Journalism Network founded in 2020 — before the inflection point generated by ChatGPT. Cases from Singapore, Japan, India, New Zealand and Austria complete a map where institutionalization is advancing in markets with specific conditions of size, resources and organizational will.
Latin America presents a more nuanced picture. Brazil leads clearly, with dedicated roles at Folha de S.Paulo and Grupo Globo of verifiable weight and track record. Argentina, Colombia and Mexico have a presence, but as exceptional cases. Active experimentation exists but institutionalization is still lagging. And the gap between the two does not close on its own — it requires deliberate organizational decisions, training models the region does not yet have consolidated, and an understanding of the transformation that goes beyond tool adoption.
The generational paradox
Newsrooms undergoing this transformation coexist with a tension the survey documents indirectly but consistently. Journalists with greater experience have the editorial judgment, institutional understanding and authority to make decisions about what gets published and how. Younger profiles arrive with greater familiarity with tools, digital workflows and automation logic. The transformation requires both at the same time, and few organizations have profiles that combine them with sufficient depth.
That tension is not new in journalism. Every technology cycle has generated similar versions of the same friction. What distinguishes it in this moment is the speed — the gap between what the tools allow and what organizations know how to do with them opens faster than teams can close it. The strongest cases in the survey show organizations that managed that tension deliberately, creating spaces for joint work between different profiles rather than betting on the substitution of some by others. Reuters, the BBC and Newsquest built structures where senior editorial judgment and technological understanding operate at the same level, with mandates that overlap by design. It is an organizational model that acknowledges the paradox rather than ignoring it.
The paradox also has a structural dimension that few organizations are naming clearly. The pressure to automate is leading some media organizations to reduce the hiring of junior profiles. If a tool can do the work of an intern or trainee, the short-term logic says there is no need to hire one. But that logic ignores what interns and trainees actually produce — not just content but accumulated experience. They are the pipeline for future editors, newsroom chiefs and editorial directors. At the same time, automation is accelerating the departure of the most experienced profiles. In some cases because they choose to retire, and in others because the organization perceives them as an obstacle when they cannot adapt to the pace of technological change. That criterion ignores what those profiles concentrate — editorial judgment built over decades — and which no system can replace. Newsrooms that stop bringing in young people and let go of senior staff believe they are optimizing when in reality they are hollowing out the two extremes that sustain the editorial culture of a media organization.
A morphology in motion
The new morphology of newsrooms does not have a final structure. The survey documents 80 roles in an open process, with organizations creating positions, redefining them or merging them when the landscape changes. What remains stable in this context is not the form but the direction of the movement.
The newsrooms that advanced furthest deliberately built the capacity to orchestrate heterogeneous systems without losing sight of what defines journalism. They knew what to delegate, what to supervise and what to reserve for human judgment at each stage of the process. That criterion did not emerge from technology alone or from journalistic experience alone. It emerged from the intersection of both, and deliberately building it was the central task that distinguished them from the teams that improvised.
This landscape will keep transforming. The roles documented in this guide will be different in three years. Some will disappear, others will multiply and several that today have no name will begin to have one. What remains is the responsibility over what gets published and the judgment with which decisions are made inside a newsroom.
80 AI Roles in Newsrooms
The complete list from the survey that underpins this chapter. 80 real positions in media organizations across 24 countries, organized into four functional families and sorted by geographic region. Each entry includes the verifiable source supporting its inclusion.
Access the full list