← Back to articles PL

Nine interpretation decisions from Poland’s Chief Labour Inspector. What decides whether a contract is employment

By 25 August 2026 Poland’s National Labour Inspectorate (PIP) had published nine decisions on individual interpretations. All of them concern one provision, Article 22 § 1 of the Labour Code, which defines the employment relationship. The mechanism is new. Since 8 July 2026 a hiring entity can ask the Chief Labour Inspector (GIP) for a ruling on its model. The question is always the same. Is the described setup, usually a civil-law mandate contract (umowa zlecenia) or a B2B deal with a self-employed contractor, in fact employment? The decisions carry dates from 28 July to 17 August. Their content points to six applicants, as three firms filed two requests each. Four decisions are favourable. Five are negative. All nine were signed for the Chief Labour Inspector by the same person. Five repeat one block of case law, including Supreme Court rulings I UK 68/05, II PSKP 23/22, I PSKP 18/21, I PK 358/16 and I PKN 416/98. The other four cite different judgments, but the test is the same throughout. Two axes: subordination and personal performance of work.

The numbers alone say little. Four favourable decisions kill the thesis that GIP is shutting the door on civil-law work. The point of this set lies elsewhere.

Case number and date Cooperation model Decision
GIP-GPP.50.14.2026.2, 28 July Mandate contracts for generating sales leads for a sole trader favourable
GIP-GPP.50.6.2026.3, 29 July B2B contractors at an IT company, framework agreement plus work orders, separately project manager services favourable
GIP-GPP.50.8.2026.7, 3 August Driver taking children with disabilities to schools negative
GIP-GPP.50.12.2026.4, 3 August Temporary work agency, task app, future arrangement negative
GIP-GPP.50.8.2026.8, 5 August Carer on the same school transport negative
GIP-GPP.50.13.2026.2, 5 August The agency again, the same model, already in operation negative
GIP-GPP.50.9.2026.6, 10 August Drivers delivering bread to a bakery’s own shops favourable
GIP-GPP.50.11.2026.3, 11 August Cooperation agreements with self-employed sole traders favourable
GIP-GPP.50.18.2026.2, 17 August The bakery again, students as shop assistants and production workers negative

One bakery, two opposite outcomes in a week

The best material in the whole set is a pair of decisions for one firm. A week apart. With opposite results.

The drivers delivering bread kept their mandate contracts. Each contract covered a single delivery run. The driver picks up the keys at the gatehouse. He gets a route. He loads at the ramp, serves five or six shops, comes back, returns the crates and the van. Pay is fixed per run and does not depend on time. The contract lets him hand the run to someone else. GIP found two features of employment missing. No duty of personal work. No subordination as to time and place.

The students at the same bakery did not keep theirs. And on paper they had more freedom than the drivers. They could vanish for weeks during exams and nothing happened. They could work in the morning at opening, go to class, then come back for a few hours. The schedule was built from what they declared.

GIP found that too little. And it split two spheres. One is the freedom to declare when I can work. The other is the freedom to decide what I do once I am at work. The first sphere was left to the students. The second stayed with the bakery, because the scope of a shop assistant’s tasks was set each time by the shop, and the scope of production tasks was set by the plan for that day.

One sentence from the reasoning is worth remembering, in our translation:

The student thus independently indicates the time window in which he is available, while the content of the work performed during that time is specified by the entity organising the sales or production process.

GIP called the freedom of entry a feature of civil-law work. But it denied it decisive weight.

The conclusion is uncomfortable. The right to refuse a task means little if, once the contractor turns up, the shift supervisor tells him what to do today.

Four rules from the full set

Rule one. Simple physical work is practically out of reach for mandate contracts. In three decisions GIP cited Supreme Court ruling I PK 126/19. It says that so called autonomous subordination does not fit simple physical work. That closes the road in the warehouse, in production, at the till, at shelf stocking, at cleaning and at stocktaking. Three of the five negative decisions concern exactly such work.

Rule two. A substitution clause must work, not just sit on paper. In the agency case the platform rules did provide for a substitute. But the mechanism was different. The old contract ended, and a new contract was signed with a new person. The substitute had no guarantee of getting the task at all. GIP held the clause to be a sham. Substitution in civil law means someone else performing the same contract, not tearing it up and signing a new one. In a favourable decision GIP cited ruling I PK 8/10 in turn. The substitute must be possible for real, not only in the contract.

Rule three. Risk on cancellation counts against the firm. The agency paid 50 percent when its client cancelled a task before the start. It paid 100 percent when the client cancelled an hour before the start or later. For GIP that is risk typical of an employer. The clauses that worked the other way, and defended the B2B models, were these. No pay for mere readiness. Contractual penalties for delay. Mandatory liability insurance. Acceptance of the deliverable. Equipment leased for a monthly rent, not handed over for free.

Rule four. Limits set by statute do not cancel employment features. A unit taking children to school as a municipal duty defended itself with two arguments. The hours of the runs follow the school timetable, not its orders. And the limits on substitution follow from the duty to vet candidates in the Register of Sexual Offenders. GIP answered that what counts is the effect, not the source. The time was fixed in advance and the contractor could not change it. The work was personal in practice. Arguments about seasonality and about careful spending of public funds were dismissed as beside the point.

Temporary agency work has its own benchmark

The two agency decisions go beyond Article 22 § 1. Following Supreme Court resolution I UZP 6/11, GIP repeated that the benchmark here is not an ordinary employment contract but a temporary employment contract. Subordination is tested against the user undertaking, that is the agency’s client. The short life of a task is no defence. Short life is what defines temporary work.

To this GIP added Directive 2008/104/EC. It provides only for employment within temporary agency work. GIP added a practical point too. If mandate contracts were allowed for one day tasks, the model would spread across the market and protection would fall.

On this approach a platform model inside a temporary work agency cannot hold. No matter how many tasks there are and how they are handed out.

The weight has moved from the contract to the practice

The thesis that PIP is tightening the screws on mandate contracts does not hold against four favourable decisions. The change is different. GIP does not ask what the parties wrote into the contract. It asks what happens on the day of work. Freedom to refuse, a loose schedule, breaks in cooperation. All of that GIP noted and treated as secondary. What decided was the moment the contractor walks in and hears what to do today.

For a firm built on mandate contracts or B2B this reverses the order of checking. First a talk with the shift supervisor, then a read of the contract. If the scope of tasks grows day by day out of the needs of the shop, the floor or the project, no clause will outweigh it. If the substitution clause has never worked and nobody knows how it would, in a dispute it will hurt rather than help. And if you pay any part of the fee for a cancelled task, you are carrying the employer’s risk.

What comes next, we will see. It is open how these decisions will play out in inspections and in the judgments of regional courts. The first requests came from firms that wanted to know where they stand. In five cases out of nine the answer went against their own position. And a copy of every decision, favourable or not, went to the Social Insurance Institution (ZUS) and the National Revenue Administration (KAS). Which is why the request to GIP is itself a decision worth calculating separately.

The above text is for information purposes and does not constitute legal advice in an individual matter. Assessment of specific cooperation requires an analysis of the contract’s content and the actual conditions of its performance.

Check in three minutes whether your mandate and B2B contracts would pass the test GIP applies.

Book a consultation →

Related Articles