The United Kingdom and 17 allied nations have jointly issued cybersecurity guidelines for engineers engaged in developing novel AI technologies. This initiative follows a recent global conference at Bletchley Park and underscores the government’s proactive stance in shaping discussions on AI security.
The latest security directives for AI engineers have been unveiled by the National Cybersecurity Centre (NCSC) of the UK. These guidelines, as per the NCSC, are geared towards enhancing cybersecurity measures for artificial intelligence and promoting secure design, development, and deployment practices.
During an event organized by the NCSC, attended by representatives from various industries and the public sector, these recommendations are set to be officially launched today.
Fresh security directives for advancing AI have been introduced.
In collaboration with industry experts and 21 international organizations and ministries worldwide, the NCSC and the US Cybersecurity and Infrastructure Security Agency (CISA) have jointly formulated guidelines to foster the development of secure AI systems.
The NCSC asserts that these guidelines will aid developers of AI-based systems in making informed security decisions throughout the development lifecycle. This includes systems developed from the ground up as well as those leveraging existing tools and services.
The primary objective is to guide engineers in embracing a “secure by design” approach when creating AI systems, ensuring that security is an integral part of new models.
NCSC CEO Lindy Cameron emphasizes the need for coordinated global efforts by governments and industries to keep pace with the rapid evolution of AI. She states, “These guidelines are a significant step towards establishing a comprehensive international consensus on digital risks and preventive measures concerning AI, emphasizing the importance of integrating security as a fundamental requirement rather than an afterthought in development.”
The guidelines are structured into four key categories: safe design, safe deployment, safe operation, each offering recommended actions to bolster security measures.
According to CISA Director Jen Easterly, these guidelines represent “a pivotal stride in our shared responsibility – embraced by governments worldwide – to ensure the creation and adoption of AI capabilities with inherent security measures.”
Easterly further underscores the critical juncture in technological advancement, stressing the imperative for global unity in promoting secure design principles and fostering resilient foundations for AI systems worldwide.
This collaborative effort highlights the significance of international cooperation in safeguarding our technological advancements and underscores the commitment to protecting critical infrastructure.
UK Spearheads the Dialogue on AI Security.
In addition to the UK and the US, countries such as Germany, France, and South Korea have endorsed these recommendations, building upon the outcomes of the global AI safety conference hosted by the UK government at Bletchley Park. The conference saw participation from top government officials, AI laboratories, and technology vendors worldwide.
The signing of the Bletchley Declaration at the event signifies a strong commitment to collaborative efforts in AI security. Leading developers like OpenAI and Anthropic have agreed to submit their cutting-edge AI models for review by the newly established AI safety institute in the UK, a pioneering initiative globally. A similar entity is also being set up by the US government.
Technology Minister Michelle Donelan commended the UK’s leadership in promoting the safe utilization of AI. With the introduction of these new guidelines by the NCSC, cybersecurity will be paramount in the advancement of AI at every developmental phase, ensuring comprehensive risk mitigation strategies are in place.
