A PwC Hungary survey conducted in May 2026 shows that a large share of workplace AI use occurs outside employers’ oversight. Among digitally active internet users surveyed, 67% of those who use AI at work reported regularly or occasionally relying on personal, free or otherwise non‑company tools. Additionally, 42% said their workplace either lacks internal rules on AI use or they are unaware of any such policy.
This so‑called "shadow AI" is risky because, while employees can save time and effort, organisations may lose control over data, business information and workflows. PwC concludes that the objective should not be to ban AI use but to create a governance environment with approved tools, clear rules and role‑specific training that enables safe adoption.
What employees use AI for and how much time they save
The survey found that 47% of digitally active internet users use some form of AI application in daily life. Usage rates differ by generation: 56–57% among Gen Z and Millennials (Y), and 35% among baby boomers.
Of those who use AI at work, 42% also use the same tools for personal purposes; this figure rises to 52% among respondents with higher education. More than 78% of workplace AI users reported at least a 10% time saving on tasks where AI was applied. Specifically, 38% reported savings of 30% or more, and another 40% reported savings between 10% and 30%. These freed capacities create opportunities to shift effort toward more complex, higher‑value work.
Most common uses and limited structured integration
Seventy‑eight percent of AI users primarily apply the technology for answering questions and quick information retrieval. Other common uses include translation (47%), image generation (35%), content writing (34%) and data analysis (31%). In workplace contexts, information retrieval is even more dominant: 73% cited it, and 35% said they would retain only that single AI function if limited to one in the next six months.
Deeper, workflow‑integrated use remains uncommon: 70% of those using AI at work adapt the technology to the specific task and do not rely on pre‑made prompts or templates. Only 12% regularly use prompts or templates for recurring tasks, 10% have partially integrated AI into standard workflows, and team‑level organised AI use applies to just 3% of respondents.
Reasons for personal tool use and consequences
Among employees using non‑corporate AI tools, 32% do so out of habit and 23% because they find them faster or more convenient. Only 22% know organisational policies well. When unsure whether content can be shared with an AI tool, 32% anonymise information, 30% create fictitious examples, 25% refrain from using AI, and only 9% select an approved corporate tool.
PwC warns that anonymisation does not fully protect sensitive business material: business plans, financial data or other confidential information can still end up in systems not controlled by the employer. In uncertain cases, just 3% of respondents turn to a manager or relevant expert, while 23% decide based on their own judgement—leading to inconsistent, hard‑to‑verify practices that may create legal, financial and reputational exposure.
Regulation, compliance and legal risk
Dr. Szűcs László, PwC Legal lawyer at Réti, Várszagi és Társai Ügyvédi Iroda, highlights that shadow AI can gradually erode employer control over organisational information and decisions. The EU AI Regulation assumes organisations must know the AI systems they use and operate appropriate governance and risk management mechanisms, including documentation when necessary. If employees widely use tools unknown to employers, meeting these obligations becomes significantly harder—especially in sectors with strict data‑handling or compliance requirements.
Training gaps and learning preferences
Eighty‑six percent of AI users had not taken any AI‑related training in the year before the survey. Among those who did complete a training programme, 93% found it useful and 24% immediately started applying what they had learned.
Of the training participants, 65% completed basic courses and 11% took advanced programmes. Respondents most commonly missed practical exercises (47%), job‑specific examples (40%), integration with daily tools (37%) and advanced tips or prompt templates (36%).
Access to corporate AI tools is limited: only 27% of employees receive such tools from their employer, while 54% use AI without organisational guidance. Twelve percent reported their employer encourages using in‑house or free solutions, and 7% said AI use is explicitly prohibited.
Learners’ preferences skew toward short, self‑contained micro‑learning units: 38% would prefer such formats. PwC experts—including Cserjes‑Kopándi Ildikó and Kelemen Ákos—stress that training delivers real change only when it is job‑specific, practical, paired with approved tools and supported by clear usage rules.
Recommendations and conclusions
PwC recommends that organisations align the following elements to convert individual productivity gains into organisational value:
- a unified AI governance framework covering employee‑initiated tools as well as corporate systems;
- availability of approved, enterprise‑level AI tools;
- clear, easily accessible usage policies and guidance;
- role‑tailored, practical training and short micro‑learning modules;
- ongoing skills assessments and regular updates of practical guidance.
The survey was an online study conducted by PwC Hungary in May 2026 among digitally affine internet users. The full dataset comprised 2,896 respondents; the base for individual questions varied depending on AI usage, work situation and training participation.
In sum, PwC’s findings indicate that AI already delivers measurable time savings and is widely used for routine tasks, yet corporate control, governance and training often lag behind. Addressing these gaps requires comprehensive governance, approved tools and targeted, practical training so that individual efficiency improvements translate into broader organisational benefits.
