datenrecht.ch

Wildhaber/Ebert: Legal opi­ni­on on “Employee par­ti­ci­pa­ti­on in the workplace”

Prof. Dr. Isa­bel­le Wild­ha­ber and Dr. Isa­bel Ebert have writ­ten a legal opi­ni­on on ADM systems (for “Auto­ma­ted Decis­i­on-Making”) in the work­place on behalf of the syn­d­icom trade uni­on and with the pro­ject manage­ment of Algo­rith­m­Watch CH, which available here is. The NZZ has repor­ted on this.

The report examines

  • the cur­rent legal frame­work in Switz­er­land and inter­na­tio­nal­ly at EU level (GDPR, Direc­ti­ve 89/391 on the pro­tec­tion of workers’ health), the Coun­cil of Euro­pe (ECHR 108, ECHR), the UN (UN Covenants I and II, UN Sus­tainable Deve­lo­p­ment Goals, UN Gui­ding Prin­ci­ples on Busi­ness and Human Rights), the ILO (ILO Con­ven­ti­ons 155 and 187), the OECD (Gui­de­lines for Mul­ti­na­tio­nal Enter­pri­ses, Recom­men­da­ti­ons on the Use of Arti­fi­ci­al Intel­li­gence) for the use of ADP systems, the regu­la­ti­on of the par­ti­ci­pa­ti­on of employees and their repre­sen­ta­ti­ves and the rele­vant court cases,
  • and the legal need for action when using ADM systems.

The experts dia­gno­se Gaps in particular

  • at Par­ti­ci­pa­ti­on Act (MitwG), which is not suf­fi­ci­ent­ly known, does not pro­vi­de for sanc­tions, does not pre­vent ter­mi­na­ti­on for eco­no­mic rea­sons (this is appar­ent­ly seen as a loopho­le in the MitwG) and does not suf­fi­ci­ent­ly cla­ri­fy when ADM systems are health-rela­ted (and thus sub­ject to participation);
  • at the Invol­vement of the trade uni­onswho do not make suf­fi­ci­ent use of coll­ec­ti­ve bar­gai­ning instruments;
  • at the indi­vi­du­al enforce­ment of rightsbecau­se tho­se affec­ted “in the con­text of sur­veil­lan­ce and dis­cri­mi­na­ti­on” with the “struc­tu­ral-syste­mic” effects of ADM systems are often not indi­vi­du­al­ly iden­ti­fia­ble and both sub­stan­ti­ve and pro­ce­du­ral hurd­les have to be overcome;
  • at the coll­ec­ti­ve law enforce­ment ex postbecau­se the FADP is desi­gned for indi­vi­du­al enforce­ment despi­te the FDPIC’s aut­ho­ri­ty to issue orders, labor inspec­to­ra­tes often only inter­ve­ne when ADM systems have harmful effects on health and pro­ce­du­ral bar­riers stand in the way.

To reme­dy at least cer­tain gaps, the report postu­la­tes (the NZZ wri­tes some­what mali­cious­ly: “as desi­red”) the fol­lo­wing mea­su­res at legis­la­ti­ve levelThe experts make con­cre­te sug­ge­sti­ons for imple­men­ta­ti­on in each case:

  • A Streng­thening of rights of employee repre­sen­ta­ti­ves and asso­cia­ti­ons. In par­ti­cu­lar, employers must be obli­ged to inform employees not only indi­vi­du­al­ly in accordance with the DPA, but also coll­ec­tively. When using ADM systems and in the event of sub­se­quent chan­ges that may have nega­ti­ve effects, infor­ma­ti­on and con­sul­ta­ti­on rights must be ensured;
  • the Objec­tion opti­ons of employees and their repre­sen­ta­ti­on in ADM systems should be improved;
  • the­re are struc­tures for Super­vi­si­on and con­trol for exam­p­le in the area of risk manage­ment or through impact assessments;
  • Solu­ti­ons with the Social part­ner­ship should be drawn up befo­re any revi­si­on efforts are con­side­red at a legal level.

At com­pa­ny level, a Duty of care employers in accordance with the UN Gui­ding Prin­ci­ples on Busi­ness and Human Rights (UNG­Ps) that requi­re stake­hol­der invol­vement as part of due diligence.

The expert opi­ni­on Empi­ri­cal results from the Natio­nal Sci­ence Foun­da­ti­on stu­dy “Big Data or Big Brot­her? – Big Data HR Con­trol Prac­ti­ces and Employee Trust” (March 2017 to Febru­ary 2021) is pro­gres­sing. Howe­ver, one won­ders to what ext­ent the results of the stu­dy can actual­ly be used as a basis for the report. This stu­dy appar­ent­ly reve­a­led an increa­se in peo­p­le ana­ly­tics tools. Howe­ver, it even inclu­ded online satis­fac­tion sur­veys or com­pu­ter-based exit sur­veys or soft­ware for con­trol­ling video came­ras, i.e. basi­cal­ly harm­less mea­su­res. Accor­ding to the Dis­ser­ta­ti­on by Gabri­el Kas­per were the most popu­lar appli­ca­ti­ons of Peo­p­le Ana­ly­tics, in that order:

  • web-based satis­fac­tion surveys
  • Video sur­veil­lan­ce systems
  • Exit sur­veys
  • RFID bad­ges
  • Feed­back instrument

In con­trast, robots are curr­ent­ly hard­ly used for recruit­ment or sen­ti­ment ana­ly­sis. This alo­ne hard­ly indi­ca­tes a need for action. The report does not quan­ti­fy how high the pro­por­ti­on of actu­al ADM systems within the peo­p­le ana­ly­tics tools was in each case. Howe­ver, Gabri­el Kas­per sta­ted that auto­ma­ted indi­vi­du­al decis­i­ons within the mea­ning of the FADP hard­ly ever occur in peo­p­le analytics.

In this respect, we can only con­clude that the pro­blem descri­bed in the stu­dy is poor­ly docu­men­ted despi­te the stu­dy at the time. The sug­ge­sti­ons for impro­ve­ment can the­r­e­fo­re cer­tain­ly clo­se gaps, but whe­ther the­re is a real need for this remains to be seen.

Howe­ver, more sen­si­ti­ve ADM tools are likely to incre­a­sing­ly use AI systems. Per­haps we should the­r­e­fo­re wait for the dis­cus­sion on the regu­la­ti­on of AI befo­re reac­ting to pro­blems in the sector.

Aut­ho­ri­ty

Area

Topics

Rela­ted articles

Sub­scri­be

Sub­scri­be to news →