Artificial intelligence continues to evolve at a rapid pace, presenting both extraordinary opportunities and complex challenges related to its management and accountability.
What happened
Recently, Google showcased the capabilities of its new “anything-to-anything” generative AI model, capable of manipulating and creating multimedia content with unprecedented flexibility. This includes the ability to generate extremely realistic deepfakes, as documented by The Verge AI. The ability to transform an object into an animated video or alter complex scenarios raises questions about the spread of misinformation and the manipulation of reality.
In parallel, AI adoption in the enterprise sector continues, with concrete examples of application. Ferrari, in collaboration with IBM, is using artificial intelligence to revolutionize the experience of its Formula 1 fans, creating “superfans” through personalized interactions and targeted content, as reported by TechCrunch AI. This demonstrates AI's potential to enhance customer engagement and loyalty.
However, as AI's complexity and integration into products and services increase, a fundamental debate intensifies: who owns “correctness” in an organization's AI products? An article on Alokit via Hacker News AI filtered highlights how there often lacks a clear attribution of responsibility for the validity, accuracy, and ethical impact of AI systems, leaving a potentially dangerous governance gap.
Why it matters
The rapid evolution of generative AI, capable of creating content indistinguishable from reality, poses significant challenges to public trust and the veracity of information. The proliferation of deepfakes and the ease with which they can be produced require not only advanced detection tools but, more importantly, a robust ethical and legal framework to limit their abuse. The ability to manipulate perception can have profound repercussions on society, from political disinformation to fraud.
The application of AI in sectors like sports entertainment, while promising for engagement, must be conducted with transparency and respect for user privacy. Excessive personalization, if not managed ethically, can lead to information bubbles or manipulation of preferences, undermining individual autonomy. This highlights the challenges of AI in business.
The crucial point raised by the lack of “ownership” over AI correctness is a wake-up call for all companies developing or implementing intelligent solutions. Without clear attribution of responsibility, the risk of bias, systematic errors, or unfair decisions by AI increases exponentially. This is not just a technical problem, but a structural gap in AI governance that can erode consumer trust and lead to serious legal and ethical consequences.
The HDAI perspective
For Human Driven AI, technological advancement must always be balanced by meticulous attention to human and social impact. Recent developments demonstrate that the power of generative AI requires an ethic of responsibility intrinsic to every phase of its development and implementation. Creating ethical AI necessarily involves defining clear roles and responsibilities within the organizations that produce it. It is not enough to develop powerful models; it is crucial to establish who is responsible for their reliability, fairness, and the prevention of harmful uses. This approach is vital for building AI systems that serve humanity and not become a source of new risks. These themes will also be central to the HDAI Summit 2026, where experts will discuss how to translate these principles into concrete practices. This focus aligns with the broader goal of fostering an Italy AI summit that champions responsible innovation.
What to watch
It will be crucial to observe how companies respond to this growing need for clarity in internal AI governance. The introduction of roles like AI Ethics Officers or the creation of cross-functional committees dedicated to model correctness could become standard practice. Concurrently, emerging regulations, such as the EU AI Act, will need to evolve to address the challenges posed by “anything-to-anything” models and ensure that responsibility is not an abstract concept but a concrete and verifiable obligation.