Bonifacio VIII
A disenchantment device against AI safety-washing
Today we are witnessing a dangerous privatization of digital security under the guise of corporate “safety-washing.” Closed commercial chatbots use superficial interface guardrails to reassure regulators and pacify public debate, while underlying model capabilities remain unaccountable.
This initiative launches Bonifacio VIII, an inspectable, open-weights language model stripped of cosmetic commercial safety filters (guardrails). The Italian campaign releases Bonifacio VIII as a pedagogical stress-test to demonstrate how easily corporate ‘safety-washing’ barriers can be bypassed, giving researchers, activists, and citizens direct visibility into the uncensored mechanisms of generative AI models.
Even though our campaign targets algorithmic censorship policies across all of Europe, our physical model demonstrations and public stress-tests are currently focused inside Rome and Milan.
Italian Equivalent Resources
Since our physical workshops and launch events are based in Italy, you can access the core presentation slides directly:
- Bonifacio VIII Presentation PDF: Slide presentation outlining the concept and critique of commercial AI alignment.