Adrián Gonzalez Sanchez - Navigating the New Era of AI Governance
Keynote Speech
In an energetic and insightful session held shortly after lunch—when, as Adrián Gonzalez Sanchez reminded us, the audience’s energy can dip—attendees were guided through the rapidly evolving landscape of AI accountability and governance. Comparing today’s environment to the hit TV show Lost, Adrián emphasized our collective need to “catch up” on what has happened since last year, when Artificial Intelligence was already a key topic but less tightly regulated.
From Excitement to Accountability
A year ago, enthusiasm for generative AI and large language models was at an all-time high. The focus was on how AI could amplify both beneficial information and potential misinformation. Organizations saw AI as a catalyst for growth and transformation, but the pressing question remained: How do we adopt it responsibly? Now, “accountable AI” has stepped into the spotlight. Regulations have been introduced worldwide, compelling companies to move from purely ethical aspirations to verifiable compliance.
Assembling the AI Governance Puzzle
Adrián used a puzzle analogy to break down the multifaceted challenge of AI governance. Key ‘pieces’ include:
- Regulatory Requirements – Global regulations like the European Union’s AI Act, along with data privacy laws such as GDPR and CCPA, are shaping adoption.
- Industry & International Standards – Frameworks such as ISO and CENELEC offer best practices for risk management, data quality, and AI life cycles.
- National Frameworks – Various geographies introduce their own AI legislation, complicating compliance for multinational companies.
- Internal Policies & Tools – Many organizations already have assets (e.g., risk assessments, impact checklists) that can be incorporated into broader governance systems.
Products, Services, and Ownership
A recurring question is what exactly needs to be certified: the AI model itself or the end product or service incorporating it? Companies frequently rely on external cloud models from providers like OpenAI or Google, making it crucial to clarify responsibilities among adopters, providers, and other stakeholders. As Adrián noted, this role distribution is essential for achieving compliance and mitigating risk.
Practical Tools and First Steps
To simplify governance, Adrián recommended beginning with small, tangible actions. Converting regulatory requirements into technical formats—like CSVs or JSON files—helps track transparency obligations. Further, leveraging standardized impact assessments from organizations such as UNESCO or Microsoft ensures a structured approach to evaluating ethical and societal implications.
Looking Ahead
Though the regulatory environment can seem daunting, Adrián pointed to encouraging developments like Mistral AI, the Linux Foundation’s SPDX AI project, and sandbox initiatives that foster responsible innovation. Step by step, companies can assemble the pieces of this puzzle, moving from feeling “lost” to implementing a structured, robust governance framework.
“By the end of the day, even if this looks pretty complex,” Adrián concluded, “we are less Lost.” As the AI ecosystem evolves, forging the right alliances between business, technology, and regulatory bodies has never been more crucial for building trust and ensuring AI remains a positive force for us all.
