In order to protect artificial intelligence (AI) systems from rogue actors, the United States and 17 other nations presented an international agreement that requires providers to adhere to “secure by design principles.”

The Cybersecurity and Infrastructure Security Agency of the United States Department of Homeland Security and the National Security Centre of the United Kingdom jointly released a 20-page document on Sunday that offers a set of guidelines to ensure AI systems are designed to “function as intended” without disclosing confidential information to unauthorized users.

The guidelines, which are divided into four categories, include suggestions that can be followed at every stage of the process, from the design and development of AI systems to their deployment and upkeep. The guidelines cover topics such as risks to AI systems, safeguarding AI-related assets, including models and data, releasing AI systems responsibly, and the need to keep an eye on these systems once they are made public.

Australia, Canada, Chile, the Czech Republic, Estonia, Germany, Israel, Italy, Japan, Nigeria, Poland, and Singapore are among the other nations included in the pact.

The U.S. Cybersecurity and Infrastructure Agency’s Jen Easterly told Reuters that although the agreement is nonbinding and contains more broad advice for AI providers, the rules show how important it is to put AI system safety first.

“This is the first time that we have seen an affirmation that these capabilities should not just be about cool features and how quickly we can get them to market or how we can compete to drive down costs,” Easterly told the news outlet.

Tech companies have been racing to implement AI systems more quickly in the last year, but some policymakers and industry leaders are worried about the possibility of unchecked AI growth.

The Biden administration last month issued an executive order aimed at controlling the hazards associated with artificial intelligence. In addition to asking federal agencies to expedite the development of methods so AI systems may be trained while maintaining the privacy of training data, the order incorporates additional safety standards and worker protection principles.

The order will also evaluate how agencies collect and use commercially available information with personal data, as well as introduce a tool to expand grants for AI research in areas such as health care and climate change.