Canada’s federal government has co-signed a new AI standards pledge with a handful of other countries and major AI developers.
The news: The Government of Canada revealed on Monday that it has backed a fresh international declaration on AI that is geared towards establishing “the necessary foundations for safe adoption.” Leaders from five other countries (Bahrain, Finland, France, Norway, and the Netherlands) and 10 AI companies (including Anthropic, Mistral, OpenAI, and Canadian champion Cohere) have also agreed to the “Paris call for AI safety and security.”
The firms have committed to permitting some independent testing of their models, providing early warnings, and shutting down their systems when things go awry “to prevent cascading systemic risks and to formulate collective responses to emerging threats.”
From the source: “Frontier models transcend national borders in their development, deployment and impact, and their opportunities and risks are therefore inherently global,” the joint statement reads. “We all have a role in protecting our societies while enabling research, innovation and the wider use of AI … No actor, be it a state, a company, or an international organization, can solely manage all these risks.”
The context: Political and industry leaders have become increasingly concerned about the risks associated with AI as rogue agents have attempted, sometimes successfully, to hack other businesses and governments. A decade ago, large language models (LLMs) struggled to string together a sentence; in recent months, AI systems have executed a decade’s worth of progress in mathematics and shown their capacity to complete sophisticated real-world plans over time.
The declaration’s signatories have agreed to disclose frontier model capabilities, limitations, and risks, develop and adhere to common technical standards, and closely monitor and report on these systems to minimize potential harm. Canadian AI leaders like Geoffrey Hinton and Yoshua Bengio have been calling for measures like this to be implemented.
Final thought: The United States and Chinese governments, as well as other leading AI firms like Google and Meta, have not signed the declaration, which remains open to new endorsements. It is not yet clear how these commitments will be enforced, or the extent to which they will be effective without broader buy-in, especially from the two nations leading the global AI race.
Feature image courtesy Unsplash. Photo by Aleksandr Galenko.