The hazard of industry influence on the independent scientific panel

Killian Gross
European Commission DG CNECT A2

24 September 2024

Hazard of industry influence
on the independent scientific panel

Dear Mr. Gross,

1. We write to emphasise the need for careful scrutiny of all candidates for the scientific panel to be established for under the EU AI Act.[1] Article 68 (2)(b) makes it clear that experts may only qualify for the panel if they can demonstrate “independence from any provider of AI systems or general-purpose AI models” (“AI providers” hereafter). This is repeated in point (c), and in the fourth paragraph of that Article.
   
   
2. The panel’s role will be important. It will support the enforcement of the EU AI Act at Member State and Union level, aid in the classification of systemic risks from general purpose AI systems,[2] and may examine documentation from AI providers.[3]
   
   
3. With such critical responsibilities and influence, the need for “impartiality and objectivity”[4] cannot be overemphasised. Decisions will be successfully challenged by AI providers if the panel is not objectively impartial, as occurred most recently in a pharmaceuticals case in March.[5] It is essential that the Commission avoid any appearance of conflict of interest in view of the global profile and legislative significance of the AI Act.
   
   
4. AI Providers have invested heavily over many years to establish think tanks and capture academia as a means of indirectly influencing enforcers.[6] In such an environment, it will be impossible for the Commission to “actively manage”[7] conflicts of interests in the scientific panel.
   
   
5. As noted by David Michaels, the longest serving Assistant Secretary of Labor for the Occupational Safety and Health Administration, “conflict of interest cannot be managed. It must be eliminated.”[8] We urge the Commission heed to this advice and not repeat the same mistakes as other sectors.
   
   
6. Speaking earlier this month, U.S. Assistant Attorney General Kanter worryingly observed a diverse set of domains where “conflicts of interest and capture have become so rampant and commonplace that it is increasingly rare to encounter a truly neutral academic expert.”[9] A former Chief Economist of the European Commission made the same point this month.[10]
   
   
7. The Commission is fortunate that along with Member States it funds the best academics in their fields, and will not find it difficult to establish an excellent and impartial scientific panel.[11]

Sincerely,

Dr Kris Shrishak
Senior Fellow, Enforce

Dr Johnny Ryan
Director, Enforce