Joana Pinto argues that attacks by AI agents “are still a crime”

News6 October 2026

Joana Pinto, a partner in the Digital Contracting & Compliance practice group at Antas da Cunha Ecija, spoke to Observador about accountability for cyber-attacks carried out by artificial intelligence agents.

The lawyer argues that an attack carried out by an agent “is still a crime simply because AI was used” and that, even without a specific regulatory framework for AI, unauthorised access to a computer system remains a punishable offence. She warns that liability cannot be left in a “no-man’s-land”, whilst acknowledging the difficulty of proving the elements of liability, given the complexity of the entities involved in these systems.

She also points out that the AI Act already sets out obligations for providers of models posing systemic risk, such as assessments and testing, risk mitigation and the reporting of serious incidents, which may be relevant in determining whether appropriate safeguards have been put in place.

Read the full article in Observador

La imagen presenta un diseño abstracto con líneas en tonos azules y verdes que crean un efecto de profundidad.

Related professionals

LATEST FROM #ECIJA