The US, UK, and several other countries have jointly signed an agreement aimed at ensuring the security of AI technology. A senior United States official revealed on Sunday that this detailed agreement is designed to assist companies in creating AI models that are ‘secure by design.’
The comprehensive 20-page document unveiled that eight global companies have agreed to develop AI models that prioritize both safety and public well-being. This non-binding agreement necessitates the surveillance of AI models to prevent abuse and ensure the protection of data against breaches or tampering.
Jen Easterly, the director of the US Cybersecurity and Infrastructure Security Agency, emphasized the significance of multiple companies collaborating to guarantee the safety of AI models. Easterly stated, ‘This is the first time we have seen an affirmation that these capabilities should not merely focus on innovative features or rapid market entry or cost competitiveness. These guidelines represent an agreement that security is the most crucial aspect to be addressed during the design phase.’
Alongside the United States and Britain, 18 other notable countries, including Germany, Italy, the Czech Republic, Estonia, Poland, Australia, Chile, Israel, Nigeria, and Singapore, have signed the agreement.
The framework deals with techniques aimed at preventing AI models from being hacked by intruders and includes recommendations such as launching the models only after rigorous testing to meet security requirements.
In Europe, efforts to draft guidelines for AI are leading the world, with France, Germany, and Italy signing agreements specifically focusing on ensuring AI safety through ‘mandatory self-regulation via codes of conduct.’ Meanwhile, the US aims to mitigate AI risks with a focus on enhancing its security.
