Information Disorder In The Age Of AI: What It Means For Employers

Misinformation and disinformation are usually treated as societal, political or geopolitical risks. In an AI-enabled information environment, they are also becoming workforce risks. As generative tools make persuasive synthetic content easier to produce and harder to verify, employers face growing challenges around reputation, employee relations, data governance, legal exposure and AI-assisted decision-making.

The old problem of new information technologies

A donkey’s body, fish-like scales, and a distorted human face: this was the ‘papal ass’, one of early modern Europe’s stranger media sensations. Supposedly discovered in the River Tiber in Rome in 1523, the creature was presented in pamphlets as a divine warning against corruption in the Catholic Church.

The creature never existed, yet the story spread rapidly, as woodcut illustrations reproduced in cheap pamphlets were carried through Europe’s expanding communication networks. This happened because the printing press had transformed the geography and pace of information: rumours that might once have remained local could now be replicated at scale, allowing sensational claims to travel faster than verification.

Five centuries later, generative AI and algorithmic distribution platforms are producing a comparable disruption. Persuasive synthetic text, images, and audio can be created cheaply and at scale, tailored for virality, and distributed globally within minutes. The technologies differ, but the institutional challenge remains strikingly familiar.

The World Economic Forum’s recent Global Risks Report 2026 identified misinformation (false information shared without the intent to deceive) and disinformation (information deliberately fabricated and spread to deceive) as the most severe global risk linked with technology. While this challenge is generally framed in democratic or geopolitical terms, this article explores a less examined implication: how misinformation and disinformation are increasingly becoming an operational risk for employers. As the boundaries between social discourse and workplace perception blur, information integrity becomes a matter of governance rather than public relations.

AI as amplifier and feedback mechanism

Artificial intelligence has accelerated and automated existing forms of misinformation, as generative systems lower the barrier to producing plausible narratives, making synthetic images and text appear authoritative and persuasive, while weakening authenticity cues. As Nina Schick showed in her book Deep Fakes and the Infocalypse, these tools are beginning to enable the industrial-scale production of synthetic media capable of reshaping information environments. The WEF Global Risks report similarly warned against the information risks linked with the rise of GenAI: lowering “the barriers for content production and distribution” can potentially enable “threat actors, state agencies, activist groups, and individuals who may or may not have criminal intentions” to “automate and expand disinformation campaigns, greatly increasing their reach and impact”. As a result, the report concludes, “within a decade, deepfakes and AI-generated misinformation could become ubiquitous, making it impossible for citizens to distinguish truth from deception”.

Large language models are trained on vast volumes of online material. Where that material is saturated with distortion, bias or fabrication, models may reproduce or amplify the same patterns. Over time, a circular dynamic can emerge: misleading AI-generated content enters the online ecosystem, that ecosystem supplies data for later models, and the feedback loop strengthens. The risk is less like a single error and more like a map repeatedly redrawn from other maps, rather than checked against the terrain: each version may look authoritative, but small distortions can become embedded until they are treated as features of reality. As synthetic content becomes cheaper to produce and harder to verify, the informational foundations on which organisations rely become less stable.

This is precisely what the American philosopher Lee McIntyre referred to as the ‘post-truth’ condition, whereby misleading narratives circulating more easily than verified knowledge can erode the shared factual ground on which institutions depend. The 2025 Edelman Trust Barometer rang the same alarm bells, noting that, according to 63% of respondents globally (and up to 75% in some markets), “it is becoming harder to tell if news is from respected media or an individual trying to deceive people”.

From societal threat to organisational variable

This logic poses a significant problem for the future of work. Even though misinformation is generally framed as a political or cultural problem, it now brings about an increasingly serious, if less often explored, set of three challenges for employers.

Reputational risk at algorithmic speed

First, reputational narratives now move at algorithmic speed. Large-scale studies of digital information diffusion show that this can spread “farther, faster, deeper and more broadly” online than in traditional communication environments, giving empirical weight to the old adage that “a lie can travel around the world and back again while the truth is lacing up its boots”. Misleading accounts of workplace conduct, safety incidents or corporate policy can circulate widely before an organisation can investigate or respond. This was famously the case in 2023, when a group of academics submitted allegations of serious wrongdoing by KPMG to an Australian parliamentary inquiry, only for these to prove entirely fabricated by Google’s Bard AI tool (now known as Gemini).

In a tight labour market, distorted employer branding can shape candidate behaviour and retention dynamics. This challenge sits alongside the broader governance pressures highlighted in our Future@Work 2026 report, with 60% of organisations identifying cybersecurity and data privacy as the biggest risk associated with AI adoption. As digital systems increasingly mediate how information about organisations circulates and is interpreted, reputational risk becomes closely intertwined with data governance and platform dynamics.

Internal cohesion and AI literacy

Second, information disorder can weaken internal cohesion. Employees encounter external narratives about regulation, diversity, economic conditions and technological displacement alongside organisational communications. When inaccurate claims about company policy or legal obligations gain traction internally, trust and clarity suffer. In this context, our Future@Work 2026 report echoes the findings of the 2025 WEF Future of Jobs report, suggesting that many organisations may be poorly equipped to manage these dynamics: 93% report workforce skills gaps linked to AI adoption, including 44% identifying a lack of AI literacy across the workforce. In environments where digital and AI literacy remain uneven, employees may struggle to assess the reliability of the information circulating around them.

Data-dependent decision-making

Third, organisational decision making is becoming increasingly data dependent: AI-enabled recruitment, workforce planning and compliance systems rely on data quality and model integrity. Contaminated inputs or biased training material can therefore translate directly into legal and operational exposure. Here again the data from Future@Work 2026 is revealing, and aligns closely with the findings of the 2026 Deloitte Human Capital Trends report: 81% of organisations report gaps in governance and risk capability linked to AI, with 39% highlighting weaknesses in data governance around quality, security and access. In practice, this means many organisations are deploying data-driven tools faster than they are developing the institutional frameworks required to oversee them effectively.

This challenge is compounded by a widely reported tendency towards ‘cognitive offloading’, whereby workers outsource mental effort to AI tools in ways that may weaken memory, problem-solving, and critical thinking over time. The Henley Centre for Leadership 2026 report Leadership Futures: Redefining Leadership in the Age of AI has warned that “what increasingly matters is not what AI produces, but how people prompt it and how critically they evaluate what it returns… As technology makes decisions easier, judgment can start to give way to convenience”.

The investment imbalance

These insights provide useful context to one of the most alarming findings of our report, namely that organisations appear to be responding asymmetrically to these challenges: 74%, in fact, plan to increase spending on technology in the coming year, while only 5% expect to spend more on workforce development than on technology. This imbalance sits uneasily alongside the fact that 64% of employers expect soft skills such as judgment and critical thinking to become more important as AI reshapes work.

Information disorder therefore exposes a broader readiness gap. Technology can accelerate analysis and decision-making, but reliability still depends on governance, sound data and people capable of questioning what systems produce.

Employment law and regulatory exposure

Information disorder rarely presents a self-contained legal problem. More often, it complicates familiar areas of exposure, including discrimination, data protection, fairness, employee speech and AI governance.

AI-assisted decisions and discrimination risk

As we discussed elsewhere, AI is increasingly used at various stages across the employee life cycle. Systems may screen candidates, allocate work, assess performance, recommend promotion or identify employees for disciplinary action or redundancy. In the UK, these issues are governed through a dispersed combination of employment, equality and data protection law rather than a single AI-specific regime.

Where employers rely on AI to make or support decisions about recruitment or management and the underlying data reflects historical inequalities or contains fabricated material, the resulting decisions could well disadvantage protected groups and contribute to creating direct or indirect discrimination risks under the Equality Act 2010. Organisations must understand how the AI system operates, monitor its outcomes, and provide transparency and meaningful human scrutiny.

Automated decisions and data protection

The Data (Use and Access) Act 2025 (DUAA) has recently widened the circumstances in which organisations may make significant decisions through solely automated processing, while retaining safeguards for affected individuals, unless special category data is involved. Meanwhile, the Information Commissioner is currently developing revised guidance and has identified automated recruitment as a regulatory priority, and will produce a statutory Code of Practice on AI and ADM as required under DUAA. Employers should therefore consider fairness, transparency, accuracy, lawful basis, bias monitoring and routes for challenging consequential decisions. A nominal human sign-off offers little protection where the reviewer lacks the information, authority or time to interrogate an output.

Grievances, evidence and employee voice

Information disorder can also complicate other workplace decisions and create legal risk. An employer relying on an AI-generated report, manipulated image, fabricated allegation or distorted dataset without testing its reliability may struggle to show that it acted fairly and reasonably. Equally, employers responding to inaccurate statements made internally or online must distinguish between deliberate falsehoods, good-faith mistakes, expressions of protected belief and disclosures that may attract whistleblowing protection. As we have explored elsewhere in relation to grievances, AI tools may also make workplace processes more difficult to manage by enabling employees to generate large volumes of apparently detailed material in support of complaints or disputes, increasing the burden on employers to separate genuine concerns from unsupported or unreliable assertions. Premature action by the employer may therefore create unfair dismissal, detriment, discrimination or whistleblowing exposure.

The EU AI Act

Outside the UK, the landscape is shifting rapidly, with the EU AI Act providing the clearest emerging example of a more integrated framework. It treats specified systems used in recruitment, selection and worker management as ‘high risk’, bringing requirements concerning risk management, data governance, documentation, traceability, accuracy and human oversight. Following the May 2026 political agreement on the AI Omnibus, these high-risk rules are scheduled to apply from 2 December 2027. Separate transparency obligations will begin to apply from August 2026, including requirements for certain AI-generated or manipulated content, such as deepfakes, to be identifiable or disclosed. For AI systems that generate or manipulate synthetic content and are already on the EU market before 2 August 2026, the obligation to machine-mark outputs as AI-generated is expected to benefit from a short grace period until 2 December 2026.

These developments make information integrity a matter of legal and enterprise governance: employers need clear ownership of AI-assisted decisions, reliable audit trails, supplier due diligence, robust investigation protocols, effective escalation routes and human reviewers who are empowered and equipped to challenge automated outputs.

The resulting question, then, is how employers can translate these principles into practice.

Data infrastructure as defensive architecture

Employers’ response to information disorder should rest on two mutually reinforcing pillars: creating resilient data infrastructure and strengthening the human capabilities needed to exercise meaningful judgment and oversight.

Workforce data as infrastructure

The first reflects a central recommendation of our Future@Work 2026 report: organisations should build robust data foundations to support sustainable, evidence-based decision-making. Workforce data should be treated as infrastructure rather than a reporting output: information used to inform recruitment, workforce planning, performance management or compliance should be consistent, accurate and traceable, with clear ownership and links to wider strategic objectives.

Provenance and supplier oversight

Provenance becomes particularly important where systems draw on synthetic content, external data or complex AI models whose outputs cannot always be fully unpacked after the event. Employers may not be able to trace every element of a model’s reasoning, especially where sophisticated third-party tools are involved. They should therefore focus on the safeguards that are realistically within their control: understanding what data sources are being used, what role the tool plays in employment decisions, what validation has been carried out, and what routes exist for challenge or escalation. Practical measures include maintaining inventories of AI tools and data sources, documenting their intended use, requiring suppliers to explain how they test for accuracy and bias, and preserving audit trails that show how consequential decisions were reviewed. Meaningful human oversight is therefore a governance challenge in its own right, rather than a simple safeguard that can be assumed once a person is placed in the process.

The same scrutiny should extend to suppliers. Procurement processes should examine how providers source and test their data, mitigate bias, monitor model performance and respond when inaccurate or manipulated information enters a system. Contractual assurances have limited value unless employers retain enough visibility and expertise to test them.

Integrated governance

The Future@Work 2026 report also recommends stronger cross-functional governance. HR, legal, data, technology and operations should share clear ownership of data quality and AI risk, supported by agreed escalation routes and board-level oversight of systems influencing consequential workforce decisions. Auditing should assess both technical performance and whether outcomes remain reliable and equitable.

These controls should form part of an ongoing strategic process rather than a one-off compliance exercise. A rolling planning cadence, supported by horizon-scanning and scenario testing, can help employers examine how systems would respond to manipulated data, fabricated content or supplier failure and update safeguards accordingly. Integrating people analytics, AI governance, privacy, cyber risk and employment law compliance also allows information failures to be identified before they become discriminatory decisions, regulatory breaches or reputational damage. Data protection principles such as transparency, accuracy and accountability are especially important here, as they help define the standards by which information should be collected, interpreted and used.

Strong infrastructure makes misinformation more visible, traceable and containable. Its effectiveness, however, depends on people who can interpret evidence, question outputs and intervene. Defensive architecture is therefore only the first pillar of organisational resilience. The second lies in developing the human capabilities required to make meaningful oversight a reality.

Human judgment as an organisational safeguard

Meaningful human oversight requires more than placing an individual at the end of an automated process, as reviewers need sufficient knowledge, authority and time to interrogate the evidence, identify gaps and depart from a system’s recommendation where appropriate.

This reinforces another recommendation of our Future@Work 2026 report: technological deployment should be matched by workforce development, thoughtful job design and human-centred capability building. The current investment imbalance suggests that many organisations remain some distance from that goal.

Resilience begins with practical AI literacy. Employees should understand how generative systems produce answers, the limitations of their training data and the possibility of hallucination, bias or manipulation. Training should help them evaluate sources, verify consequential claims and recognise when apparently credible material requires further scrutiny. Clear escalation routes should enable concerns about dubious data or AI-generated outputs to reach the right decision-makers quickly.

The report also recommends moving towards a skills-based workforce strategy that develops capabilities across roles and functions rather than confining AI expertise to technical specialists. This is especially relevant because 64% of employers expect demand for soft skills to increase as AI reshapes work. Critical thinking, judgment, communication and contextual awareness help employees interpret information within its wider legal, ethical and organisational setting. In an environment characterised by abundant content and weakened authenticity cues, these skills become part of an organisation’s risk infrastructure.

Leadership matters too. Managers should communicate clearly during uncertainty, respond quickly to inaccurate internal narratives and create a culture in which challenging an automated output is treated as responsible practice. They should also guard against cognitive offloading by clarifying which decisions AI may support, which require independent verification and where human accountability remains decisive.

Cross-functional exercises can embed these capabilities collectively. Simulations involving fabricated allegations, manipulated evidence or contaminated datasets can test whether HR, legal, data, communications and operational teams know how to verify information, escalate concerns and coordinate a response.

The goal is a workforce that can use AI confidently while retaining the capacity to question it. Technology may determine how quickly organisations generate and process information. Investment in people will determine whether they can distinguish insight from distortion and act accordingly.

Building resilience in an age of information disorder

The story of the ‘papal ass’ offers a final lesson. The printing press may have created an information environment that institutions were initially ill-equipped to govern, but societies gradually developed stronger norms of verification, accountability and editorial scrutiny. 

Today’s challenge is greater in speed, scale and technical complexity, yet the underlying principle remains the same: technological change must be accompanied by institutional adaptation. Employers cannot prevent every fabricated claim, distorted dataset or unreliable AI output from entering their organisations. They can, however, determine how far such information travels, how heavily it influences decisions and how quickly it is challenged.

By combining robust data foundations with meaningful human oversight, cross-functional governance and sustained investment in judgment and critical thinking, organisations can build resilience into the way they create, interpret and act on information. The task is substantial and urgent, but employers still have a choice: to allow AI to amplify disorder, or to build the governance, infrastructure and judgment needed to support more informed and responsible decisions.

Image details: Workshop of Lucas Cranach the Elder, Monk Calf, 1523. Woodcut on paper, image: 18.4 × 11.3 cm, sheet: 25.6 × 14 cm. Staatliche Kunstsammlungen Dresden, Kupferstich-Kabinett, A 1900–671. Details can be found here: https://artsandculture.google.com/asset/papal-ass-workshop-of-lucas-cranach-the-elder/fAE79ZFRTcIoSQ

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.