Criminologist on Hugging Face breach: multi-agent AI is the real safety frontier
sebkrier · x · 2026-09-02
Criminologist Gian Maria Campedelli reacts to OpenAI's report on the Hugging Face breach, arguing that beyond the anthropomorphized hype, the episode highlights an overlooked point: AI's future salience will come from interacting agents, not passive tools.
- Multi-agent AI systems are going mainstream, implying a qualitative and quantitative leap in safety challenges over single agents
- Social science knows little about how opinions form, goals get steered, values develop, and strategies diffuse in such systems
- His open-access paper A Criminology of Machines (Theory and Society) draws on Actor-Network Theory to argue AI agents should be treated as entities with computational, social, and legal agency, proposing a dual taxonomy of multi-agent risks
More from AGI Musings
- Hospitals Are Using AI to Raise Prices, Argues Antitrust Commentator Matt Stoller — GaryMarcus · 2026-09-03
- Cooperative AI Summer School 2026 releases talks, gathered 45 early-career researchers — xuanalogue · 2026-09-03
- AI diffusion debate: digital tech now spreads with or without human adoption — soleio · 2026-09-03
- AI agents break the internet's three-layer transaction fraud validation chain — arampell · 2026-09-03
- University instructor's field notes: students see no AI progress yet believe AI will ruin their lives — LessWrong 精选 · 2026-09-03
- Edge AI won't kill cloud CapEx: five arguments against the cloud-killer narrative — yangastas_paradise · 2026-09-03