Dr Karine Caunes, Digihumanism Executive Director and Research Associate at the Jean Moulin Lyon 3 University, was selected to participate in the drafting of the General-Purpose AI Code of Practice.

She was invited to provide expert insights in three of the four Working Groups sessions organised in the third week of November 2024:

• Monday 18 November, WG 4 Internal risk management and governance of General-purpose AI providers
  
  How can independent systemic risk assessments be adapted to the magnitude and nature of the relevant systemic risk, e.g., with regards to information security, scope of testing, time to test, transparency, etc.? (Measure 17)

• Tuesday, 19 November, WG2 Risk identification and assessment  

  Have you identified any major oversights in the draft, potential loopholes or measures that are not going to have the intended effect or even seem harmful to implement the AI Act?

• Thursday 21 November, WG4 Transparency and copyright 

  
  1/ How do you recommend that the Code of Practice specifies the obligations regarding Article 53 1b when the General-Purpose AI model distribution is implemented by means of third-party distribution channels?
  
  2/ Do you consider effective and proportionate an obligation to draw-up, keep up to date and make available to the AI Office upon request information on data sources used for training, testing and validation and on authorisations to access and use protected content for the development of a General-Purpose AI model?