General Assembly adopts landmark resolution on artificial intelligence
On 21 March, the United Nations General Assembly unanimously adopted the first global resolution on Artificial Intelligence. Resolution A/78/L.49 promotes “safe, secure and trustworthy” AI systems encouraging all UN Member States to safeguard human rights; protect personal data; and monitor AI for risks. The resolution underscores the risks posed by improper or malicious design, development, deployment, and use of AI systems, which could undermine human rights and fundamental freedoms. It recognizes the technological disparities between countries and urges support to close the digital divide and increase digital literacy.
Full article: General Assembly adopts landmark resolution on artificial intelligence
Artificial Intelligence Act: MEPs adopt landmark law
This month, the European Parliament adopted the AI Act. The law, which aims to safeguard fundamental rights and promote innovation, was approved with a majority vote. The Act highlights certain uses of AI that may infringe on citizens’ rights, which should be banned accordingly. These include untargeted facial recognition and social scoring, among others. Similarly, biometric identification is largely prohibited, with exceptions under strict requirements. The Act also provides consumers the right to launch complaints and receive specific explanations about AI decisions. Overall, the legislation represents a significant step in regulating AI technologies globally.
Full article: Artificial Intelligence Act: MEPs adopt landmark law
EU AI Office
The European AI Office within the European Commission will be a central hub for AI expertise within the EU. The Office will be essential for the implementation of the AI Act, the comprehensive legal framework aiming to ensure AI’s safety and trustworthiness. The EU AI Office will also enforce rules for general-purpose AI models and will support EU Member States’ governance bodies. The initiative to launch the EU AI Office is part of the AI innovation package launched by the European Commission in January this year alongside the “GenAI4EU” initiative.
Full article: EU AI Office
Human rights must be at the core of generative AI technologies
During his speech at Stanford University, Volker Türk, UN High Commissioner for Human Rights, emphasized the importance of integrating human rights into the development and deployment of Generative AI. While AI has the potential to solve major global issues, it also poses significant human rights challenges that are already impacting people today. Mr. Türk highlighted specific rights affected by AI, such as the right to work, non-discrimination, access to information, and privacy.
The need for a robust governance of AI with human rights as a guiding framework is priority, including responsible business conduct and human rights due diligence, to ensure AI benefits humanity while mitigating risks.
Full article: Human rights must be at the core of generative AI technologies, says Türk | OHCHR
Eight global tech companies commit to apply UNESCO’s Recommendation
Eight global tech companies have committed to UNESCO’s ethical AI framework pledging to integrate ethical principles in AI development. The agreement emphasizes human rights protection in AI design, development, and usage, marking a unique collaboration between the private sector and the UN system. The tech companies will be required to conduct due diligence for safety, identify and mitigate adverse AI effects, and perform both pre-release and post-deployment testing. The collaboration aims to foster AI that benefits the common good, reflecting a major step in responsible AI innovation.
Full article: Eight global tech companies commit to apply UNESCO’s Recommendation
OpenAI’s Sora text-to-video generator will be publicly available later this year
OpenAI’s Chief Technology Officer, Mira Murati announced that Sora will be ready to use this year. Sora is a text-to-video generator that can create realistic and imaginative scenes based on text instructions. Open AI is planning to integrate audio and allow users to edit the content produced to enhance the experience and make it more relevant for specific purposes. The technology is trained on publicly available or licensed data including content from Shutterstock. Safeguards such as watermarks will also be available to help distinguish AI-generated content from real footage.
Full article: Open AI’s Sora text-to-video generator will be publicly available later this year
