Episodes 2 and 3 - In discussion with Salomé Lannier
"AI-based detection of human exploitation"

​Episode 2 : In discussion with Salomé Lannier (Part 1)

​

• Introduction: 00:00 - 01:46

• Salomé Lannier's presentation: 01:46 - 02:46

Her research background and interests: 02:47 - 07:57
PROTEX project (Protect against exploitation): 07:58 - 10:35

• The necessity of interdisciplinary and comparative perspectives: 10:36 - 23:20

• Practical challenges of supervision by authorities and platforms: 23:20 - 28:50

• Tech perspective: AI tools detecting human trafficking: 28:51 - 35:01

• EU AI Act applicability to these tools and the consequences on "potential" victims of exploitation: 35:01 - 43:23

• What about the AI-based tools used by the platforms? The answer is in the next episode, stay tuned :)

​

In this episode, Frédérique and Yamina are discussing with Salomé Lannier, Postdoctoral Researcher at the University of Luxembourg, whose work takes part at the intersection of criminal law, digital regulation, and human exploitation.

 

Drawing on her research and the PROTEX project, this conversation explores one of the most complex challenges in today’s digital landscape: how AI systems are used to detect human trafficking and the legal and ethical risks they raise.

 

From the absence of a clear legal definition of “exploitation” to the growing use of algorithmic tools by law enforcement and platforms, the episode highlights a critical tension: tools designed to protect can also misidentify, discriminate, or even harm the very individuals they aim to support.

 

We discuss the use of AI in detecting online sexual exploitation, the risks of biased or opaque data labelling.

 

The second part of our discussion with Salomé, in which we discuss on the blurred abilities of platforms and law enforcement and limits of existing legal frameworks, follows in episode 3 below.

​​

Episode 3 : In discussion with Salomé Lannier (Part 2)

​

 

​​
 

 

 

 

​

Legal perspectives (follow-up)

​

0:34 - 1:35 The limitation of the AI Act - the DSA as addressing the consequences.

1:41 - 2:55 Interculturality - how  sex work and human trafficking differ.

3:07 - 4:11  Liability and decisions taken by the deployer.

4:35 - 8:09  The GDPR and AI Act and the multiple purposes in the scope of online content - the legal framework in the context of the online platforms.

8:20 - 11:13  The question of implementing more protection and collaboration between public-private actors ?

11:41 - 13:28  The treatment of victims - principle of non punishment for human trafficking victims & consent. 

 

Technical perspectives

 

13:56 - 18:31  The technical issues - human over the loop - research transparency.

18:32 - 21:52  The “positioning” concept.

21:53 - 24:52  Remediation of discrimination, biases - “the women question”.

25: 00 - 28:58  Ex post control over the AI system - overblocking/underblocking concepts.

29:18 - 32:51 Explainability of AI system - automated decisions in this context of human exploitation - right of explainability.

33:07 - 36:00  Legal remedy/technical remedy if the AI is not high risk - deep fakes.

36:03 - 39:00  Implementing collaboration & ethics to address those issues - digital literacy.

 

Signature questions

 

39:57 - 41:29 What female figures inspired your research?

41: 38 - 42:45 What key advice would you give to women exploring the world of AI and Law?

​

Keywords: Exploitation, Digital labour, Human trafficking, Vulnerability, AI regulatory framework, Law Enforcement. 

References:

• Link towards Salomé's research project PROTect against EXploitation (PROTEX) 

• Salomé Lannier's OrbiLu profile (list of publications)