DiPLab’s Antonio A. Casilli recently appeared on Trenta Minuti (RAI Radio 3), alongside labour law scholar Valerio De Stefano. The conversation (in Italian) was prompted by a case that illustrates, with uncomfortable clarity, how AI rhetoric is being weaponised in the workplace. The multinational InvestCloud has announced the closure of its Italian branch, initiating collective redundancy proceedings. This is allowed by law if companies are downsizing in case of poor economic performance. But in 2024 the Italian branch recorded €500,000 in net profit and €9.9 million in revenue (up 63% on the previous year). This is not a company in crisis. So, to circumvent the law, the company employed the language of automation to justify a global reorganisation that has more to do with cost-cutting than with any technological necessity. The justification invoked: AI replacing jobs, which, in their eyes, is justified and inevitable. The broader debate about AI and employment, Casilli argues, is too often reduced to a single question: will it destroy jobs or create them? The more honest answer is this: the one certain effect of AI on work is to make it more precarious. The number of data workers does not shrink as AI expands — it grows. Behind every algorithm are human beings producing data, training models, labeling images, moderating content. Invisible, contingent, and essential. When asked for the most striking story encountered in his research, Casilli mentioned Kauna Malgwi. A former content moderator, she went on to establish the Digital Rights and Mental Health Initiative Africa, supporting the mental health of data workers exposed to traumatic content. Time magazine named her among the 100 most influential figures in AI. Her story is a reminder of what the industry prefers to forget: behind every AI system, there are workers made of flesh and blood.