The introduction of disruptive technologies into society is not a new venture for humankind. The steadily increasing frequency of ‘revolutions’, suggest that we are addicted to it. Anybody with a pulse is aware that we find ourselves in an interesting pivot in human history, where the bulk of human effort has united in pursuit of an emerging and highly disruptive technology. Previous revolutions, from the wheel to the microchip, have had a profound effect on how the average human lives, but never has the ability to utilize and contribute been so immediately accessible to so many people. All advancement comes with risk, though with previous innovations, the rollout allowed time for social structures to adjust and for governments to erect the scaffolding needed for responsible public ingest. The closest analog to our current situation is the dot-com boom which was a shadow of the current AI surge, and which we famously fumbled.
There are some legislative and organizational efforts to define a code of conduct for relevant entities in the AI development pipeline. Here in the United States, our legislators are conspicuously reluctant to act. I, for one, am not going to let any entity define my moral framework – certainly not Congress – and I think it is past time for technological innovators to define the moral path forward that our government refuses to address.
Luckily, we have some inspiring examples that define ethical AI development such as the International Finance Corporation[2] and the Association for Computing Machinery[1], but it is the EU that is blazing the trail. The European Artificial Intelligence Act (EU AI) was passed in 2025[3]. Anyone developing the next revolutionary machine learning application in their garage should take notice and adjust their models accordingly. No matter the country of origin, if anyone in the EU uses or can be affected by your products, this law applies to you.
EU AI does not affect the choices of average citizens making videos of Barack Obama breakdancing or (mis)using ChatGPT as super-Google. The law does affect how AI products reach the public space. Many of the regulations defined in the Act apply to how agents interact with users, though most restrictions apply farther up in the supply chain. The groups affected most are the ‘providers’, those who facilitate the development of AI applications.
If I could sum in a few words the regulatory posture of EU AI for the garage-tinkerers to keep in mind, it would be provenance and risk assessment. Robust audibility and auditability will be crucial when moving forward as we slug through debugging systems that we barely understand, and having a solid understanding of your risk profile will help you determine what laws apply to you, and which do not. The Act defines four ‘Risk Categories’ that determine the level of regulation that applies to a machine learning enabled product. The highest classification, “unacceptable risk” results in a total ban, followed by decreasing strictness in rules as you descend through high and limited risk tiers to the ‘minimal risk’ classification which is unregulated.
As we navigate the frontier ahead, ethics in development and implementation practices will steer the industry, and ultimately, the affect it will have on society[4]. Although there are some attempts by non-authoritative bodies to help define a novel ethical framework, the U.S. government and the oligarchal industry leaders that truly define policy, are shying from enacting regulation. Please, and this applies to all business practices, do not wait on legislators to define your moral framework.
RReferences
- Association for Computing Machinery. Code of Ethics and Professional Conduct Enforcement Policy/Procedure.↩
- International Finance Corporation. Developing Artificial Intelligence Sustainably: Toward a Practical Code of Conduct for Disruptive Technologies.↩
- High-level summary of the AI Act. EU Artificial Intelligence Act.↩
- Klarin, A., Abadi, H. A., & Sharmelly, R. (2024). Professionalism in artificial intelligence: The link between technology and ethics. Systems Research and Behavioral Science, 41(4), 557–580. doi:10.1002/sres.2994↩