UAE’s G42 among firms pledging safe AI development at Seoul Summit

G42, a leading artificial intelligence company from the UAE, joined top AI firms like Google, Meta, and OpenAI in committing to safe AI practices at the AI Seoul Summit. The companies pledged to halt their most advanced systems if they can’t mitigate extreme risks. This two-day summit follows November’s AI Safety Summit in the UK and aims to establish a network of publicly backed safety institutes to further AI research and testing.

Global leaders from ten countries and the EU agreed to align their efforts on AI safety, with the British government co-hosting the event. The new network will include existing institutes in the UK, US, Japan, and Singapore.

UN Secretary-General Antonio Guterres highlighted the dual nature of AI advancements, offering life-changing opportunities but also posing significant risks like disinformation and autonomous weapons. Guterres called for universal regulations and ongoing dialogue to prevent AI from being controlled by a select few or autonomous algorithms.

The safety commitments, also signed by Amazon, Microsoft, Samsung, IBM, xAI, France’s Mistral AI, and China’s Zhipu.ai, emphasize accountable governance and public transparency. The firms agreed to publish risk assessment frameworks and stop development of models when risks are deemed severe and unmanageable.

Aiden Gomez, CEO of Cohere, underscored the industry’s growing focus on pressing concerns such as misinformation, data security, and bias since the UK summit. He stressed the importance of addressing potential risks proactively.

Governments worldwide are racing to regulate AI amid rapid technological advances impacting education, employment, copyrights, and privacy. This summit is part of broader efforts, including the UN’s first resolution on safe AI use, recent US-China high-level talks on AI, and the upcoming EU AI Act.