When we talk about technological innovation, the dominant language is one of possibilities: automation, efficiency, scalability. Legal liability almost always enters the scene later, as an unforeseen side effect.
\nThe debate on open-source AI agents \u2014 such as OpenClaw and similar systems \u2014 follows exactly this pattern: great attention to technical capabilities, very little to the foreseeable consequences of their dissemination.
\nYet European law and recent Italian case law are drawing an increasingly clear line: when a product is inherently risky, formal neutrality is no longer sufficient.<\/p>\n
For years, software has been considered a neutral tool: it executes instructions, it does not decide, it does not act autonomously. In this context, open source and \u201cas is\u201d distribution represented a sort of legal comfort zone.
\nAI agents break this paradigm. They do not merely process data; they act: they access systems, use credentials, make operational decisions that produce concrete effects on people, data, and assets of often-unaware third parties.
\nThe question, then, is no longer purely technical: can we really continue to consider the distribution of tools whose dangerous use is foreseeable as neutral?<\/p>\n
European law recognizes two distinct categories:<\/p>\n
They are not the same, but they share a central criterion: the level of safety that a person is legitimately entitled to expect, taking into account the foreseeable use \u2014 including foreseeable misuse \u2014 of the product.<\/p>\n
A recent ruling of the Italian Supreme Court (Corte di Cassazione) on liability for defective products (the vaccines case) clarifies a point that is often misunderstood: a product may be dangerous without being defective, but its dangerousness profoundly affects the producer\u2019s obligations.
\nThe Court states that:<\/p>\n
A decisive point concerns information: generic warnings or boilerplate disclaimers are not sufficient. Information must be specific, comprehensible, and functional to an informed acceptance of risk.
\nThis shifts the center of gravity of liability: not only what is produced, but how it is designed, presented, and distributed.<\/p>\n
The law has long recognized a category of inherently dangerous products: pharmaceuticals.
\nThey can cause harm, serious side effects, even death. Yet they are not prohibited.
\nWhy?
\nBecause the law does not require the absence of risk, but a prior risk\u2013benefit assessment, controlled distribution, enhanced information obligations, and \u2014 not least \u2014 a system of continuous monitoring (pharmacovigilance).
\nThe manufacturer is not liable because the drug is risky per se, but because it failed to properly assess, communicate, or manage that risk.<\/p>\n
AI agents are not drugs, but the legal parallel is far from forced.
\nLike pharmaceuticals, AI agents:<\/p>\n
The difference is that, while this model is well established for pharmaceuticals, we are only now building it for software. But the underlying logic is the same: when a product is inherently risky, freedom of distribution cannot be absolute.<\/p>\n
This evolution becomes even clearer with the Cyber Resilience Act (CRA), which transforms cybersecurity from a technical option into a legal product safety requirement (secure by design<\/i>).
\nIn this new framework, a cybersecurity vulnerability is no longer a mere incident, but a potential product defect if it undermines the user\u2019s legitimate expectations of safety.
\nAn AI agent that can be manipulated through prompt injection or foreseeable exploits due to design flaws is not merely \u201cmisconfigured\u201d: it fails to meet the expected level of security.
\nThe CRA also introduces monitoring obligations throughout the entire product lifecycle: failure to manage vulnerabilities or inertia in releasing patches becomes a source of liability.
\nHere, the analogy with pharmacovigilance becomes even stronger: it is not enough to place a product on the market that is \u201csafe at time zero.\u201d Continuous care for security is required.<\/p>\nOpen source: the end of the legal safe harbor<\/b><\/h2>\n
A widespread misconception must be clarified: open source is not a legal safe harbor.
\nEuropean law protects collaborative research and development, but this protection applies as long as the software remains outside the market.
\nOnce an open-source AI agent is integrated into a commercial product, provided as a service, or deployed in a professional operational context, the party distributing or using it assumes full responsibility for compliance, including security.
\nIt is not possible to invoke a permissive license, \u201cas is\u201d release, or the open-source origin of the code to justify harm caused to third parties.
\nIn my view, the neutrality of code ends the moment it impacts users\u2019 rights. Whoever derives economic or operational benefit from an AI agent becomes its guarantor, regardless of who wrote the source code.<\/p>\n
AI agents are not merely more advanced software. They are systems that act in complex contexts, often opaque to those who are affected by them.
\nEuropean law, Italian case law, the Cyber Resilience Act, and the evolving discipline on dangerous products converge on one point: innovation is no longer assessed solely by what it makes possible, but by what it makes foreseeably risky.
\nAs already happened with pharmaceuticals, AI is entering a phase of maturity in which technical freedom and legal responsibility can no longer be separated.
\nThe real challenge is not to stop innovation.
\nIt is to decide how far we are willing to accept the dissemination of powerful tools without taking responsibility for their foreseeable consequences.<\/p>\n","protected":false},"excerpt":{"rendered":"
Founder and Partner, CNTTV Law Firm | Lawyer specializing in privacy, AI, and banking law | Consultant for businesses, startups, and public administrations | Sales representative | Data Governance & Legal Innovation Advisor<\/p>\n","protected":false},"author":403,"featured_media":26306,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"jnews-multi-image_gallery":[],"jnews_single_post":{"format":"standard"},"jnews_primary_category":[],"jnews_social_meta":[],"footnotes":""},"categories":[1480],"tags":[794,1535,802,1537,183],"coauthors":[1586],"class_list":["post-26307","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-legal-ita-en","tag-ai-eng","tag-automation","tag-consumption","tag-innovation","tag-pos-en"],"_links":{"self":[{"href":"https:\/\/affidaty.io\/blog\/wp-json\/wp\/v2\/posts\/26307","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/affidaty.io\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/affidaty.io\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/affidaty.io\/blog\/wp-json\/wp\/v2\/users\/403"}],"replies":[{"embeddable":true,"href":"https:\/\/affidaty.io\/blog\/wp-json\/wp\/v2\/comments?post=26307"}],"version-history":[{"count":0,"href":"https:\/\/affidaty.io\/blog\/wp-json\/wp\/v2\/posts\/26307\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/affidaty.io\/blog\/wp-json\/wp\/v2\/media\/26306"}],"wp:attachment":[{"href":"https:\/\/affidaty.io\/blog\/wp-json\/wp\/v2\/media?parent=26307"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/affidaty.io\/blog\/wp-json\/wp\/v2\/categories?post=26307"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/affidaty.io\/blog\/wp-json\/wp\/v2\/tags?post=26307"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/affidaty.io\/blog\/wp-json\/wp\/v2\/coauthors?post=26307"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}