The United Nations Educational, Scientific and Cultural Organization (UNESCO) has announced that eight global technology companies, including Mastercard, Microsoft and the Lenovo Group, have signed an agreement to build more ethical Artificial Intelligence (AI) systems, integrating the values and principles of UNESCO’s Recommendation on the Ethics of AI when designing and deploying the AI systems.
According to UNESCO, the agreement was signed in Kranj, Slovenia, in February 2024 at the second UNESCO Global Forum on AI and compels the companies to fully play their role in guaranteeing human rights in the design, development, purchase, sale, and use of AI.
UNESCO said it is the first time that companies have engaged with the United Nations in this area.
In addition to Mastercard, Microsoft and the Lenovo Group, the other companies that signed the agreement are the GSM Association (GSMA), a lobby organisation that represents the interests of mobile network operators worldwide; INNIT, a Silicon Valley-based “eating technology” company; LG AI Research, the artificial intelligence research hub of South Korea’s LG Group; Salesforce, Inc., an American cloud-based software company headquartered in San Francisco, California; .and Telefonica, a Spanish multinational telecommunications company headquartered in Madrid, Spain, which is one of the largest telephone operators and mobile network providers in the world.
In the two-page agreement, titled “Commitment of Private Sector Companies to collaborate with UNESCO to build an ethical and responsible Artificial Intelligence”, the companies acknowledged that “AI technologies, including the rapidly proliferating generative AI models, are becoming more powerful and capable at a range of tasks.”
It recognized the “enormous opportunities that AI can provide to advance productivity, growth and well-being, particularly in advancing the UN Sustainable Development Goals (SDGs)” but also noted the serious risk of harm associated with the technologies when they are developed and released without appropriate guardrails.
Some of the risks identified are the widening of inequalities; the generation and spread of disinformation, hate speech and violence at scale; the risk of pervasive discrimination based on gender, race, ethnicity, and other grounds; the enabling of cyber-attacks and fraud; violations of privacy; the upending of the existing mechanisms for the protection of intellectual property rights, including the copyright concerns of authors; and the erosion of hard-won democratic norms, including human rights and fundamental freedoms.
The parties to the agreement acknowledged that AI can contribute to the transformation of development initiatives, making them more productive, inclusive and sustainable, and agreed that to seize the opportunities it offers, and to mitigate its potential threats, they must create an ecosystem of incentives, investments and ethical governance tools, both through public and private action.
They welcomed the commitments made by UNESCO’s member States in 2021 to adhere to a number of values and principles, including respect, protection and promotion of human rights, fundamental freedoms and human dignity; the flourishing of environments and ecosystems; ensuring diversity and inclusiveness; promoting peaceful, just and interconnected societies; as well as the principles of proportionality and do-no-harm, safety and security.
Other values and principles are fairness and non-discrimination; sustainability; the right to privacy and data protection; human oversight and determination; transparency and explainability; responsibility and accountability; awareness and literacy; multi-stakeholder and adaptive governance; and collaboration, which were all laid out in UNESCO’s Recommendation on the Ethics of Artificial Intelligence.
The parties noted that such values and principles are the basis of an international consensus on the design, development, and responsible use of these technologies.
They also recognized the role that frameworks like the Recommendation can play in steering developments in the right way, through policy interventions, and partnerships, in a multistakeholder fashion.
Describing UNESCO’s Recommendation on the Ethics of Artificial Intelligence, adopted unanimously by 193 member States at the 41st session of the UNESCO General Conference, as the largest global effort and the first global normative instrument of its kind, the parties agreed that the responsibility for guaranteeing and respecting human rights is a duty for all social actors, which, in the case of the design, development, purchase, sale, and use of AI, includes businesses, civil society organizations, the public sector, and higher education and research institutions, among others.
They also agreed that due diligence must be carried out in order to identify the adverse effects of AI, and timely measures to prevent, mitigate, or remedy these effects must be designed and implemented in line with domestic legislation.
The signatories stressed that this was “an ethical minimum which must be upheld” as “the software continues to evolve as it is updated post-deployment, offering new features and improvements and addressing newly identified risks, some AI models or systems continue to be subject to training or fine-tuning, or may be impacted by system updates and other adjustments to address new improvements, use scenarios, or identified risks.”
They noted the necessity of testing, adding that ex-ante assessments, which is testing before releasing a new system to the market, is essential but insufficient, which is “why in the AI field, we are conscious of the need to establish and maintain ex-post (i.e., post deployment) risk assessments and mitigation practices.”
The technology companies said they understand that they play an integral role in shaping the ethical landscape of AI and acknowledged that as companies which develop, use, purchase and sell these emerging technologies, they have a responsibility to ensure that their products meet safety standards and comply with the essential principles and values laid out by UNESCO.
They therefore made a specific commitment to collaborate with UNESCO to build ethical and responsible Artificial Intelligence.
UNESCO’s Recommendation on the Ethics of Artificial Intelligence was the world’s first, and remains its only normative framework on AI.
UNESCO said over the past two years, demonstrable progress has been made towards implementing the framework and that more than 50 countries are now actively engaged in its implementation while multilateral cooperation has increased considerably.
