In its posi­tion paper pub­lished ear­li­er this year, TEAM-NB, the Euro­pean asso­ci­a­tion of noti­fied bod­ies, warns that delays in des­ig­nat­ing cer­ti­fi­ca­tion bod­ies under the EU AI Act risk a short­age by 2 August 2027, when AI-pow­ered med­ical devices will fall under the AI Act’s high-risk scope With­out enough des­ig­nat­ed bod­ies, man­u­fac­tur­ers may strug­gle to secure the required con­for­mi­ty assess­ments and risk miss­ing the deadline. 

Des­ig­na­tion Author­i­ties and Timelines

By 2 August 2025, each Mem­ber State must estab­lish a Des­ig­nat­ing Author­i­ty (DA) to appoint noti­fied bod­ies. Slow nation­al progress on enact­ing these laws is stalling DA set­up, delay­ing the entire des­ig­na­tion process. TEAM-NB stress­es that polit­i­cal inde­ci­sion now threat­ens the industry’s com­pli­ance capac­i­ty in less than two years. 

Con­for­mi­ty Assess­ment Routes

Two routes exist: the full des­ig­na­tion under Arti­cle 30, which reduces dupli­cate doc­u­men­ta­tion but is com­plex and time-con­sum­ing, and the extend­ed des­ig­na­tion under Arti­cle 43(3), which is faster yet equal­ly rig­or­ous. TEAM-NB advo­cates for the extend­ed route but notes uncer­tain­ty about its prac­ti­cal fea­si­bil­i­ty, poten­tial­ly delay­ing bod­ies’ readiness. 

Data and Stan­dards Challenges

Noti­fied bod­ies must secure access to devel­op­ers’ train­ing and test­ing datasets and acquire inde­pen­dent val­i­da­tion data. The antic­i­pat­ed Euro­pean Health Data Space won’t be oper­a­tional in time, leav­ing a data-gap that could hin­der AI val­i­da­tion. Fur­ther­more, har­monised AI stan­dards lag behind the Act, forc­ing bod­ies to com­pare sys­tems against the “state of the art”—a far more resource-inten­sive approach. 

Inte­grat­ed Report­ing and Definitions

To avoid dupli­cat­ing paper­work, TEAM-NB calls for inte­grat­ed safe­ty report­ing between the AI Act and exist­ing med­ical-device sys­tems. Clear, agreed definitions—especially for terms like “sub­stan­tial mod­i­fi­ca­tion”—are essen­tial to man­age AI through­out its life­cy­cle and uphold patient safe­ty and fun­da­men­tal rights. 

Source: Medtech Insight (an Infor­ma product)

Accom­pa­ny­ing this sub­ject we rec­om­mend the fol­low­ing con­tent on our website

Browse our web­site by tag “Arti­fi­cial Intelligence”