Welcome to the return of Onfido’s Policy Corner, your regular briefing on key global policy updates from the world of digital identity, regulatory compliance, AI, and data privacy. The beginning of November has been an AI extravaganza with the UK-hosted AI Safety Summit, the Bletchley Declaration and the United States Executive Order on Safe, Secure and Trustworthy AI.
The global AI conversation
The overarching theme of the global political conversation is heavily focused on regulating the most powerful AI models, and concerns over the potential existential threat they pose to national and economic security, critical infrastructure, and national public health and safety.
While it’s essential to discuss the challenges posed by these high-powered models, this focus can lead to overlooking the challenges and opportunities faced by businesses and the public today. There needs to be a balance, addressing both everyday and existential regulatory needs with the same urgency.
Onfido released the results of its AI fraud survey of 1,500 business leaders this week which demonstrates that while businesses broadly recognize the challenges posed by AI (and especially generative AI) in accelerating the speed, quantity and sophistication of fraud attacks, there is a lack of consensus on where companies ought to prioritize their response to AI.
Ensuring balance in the global conversation is essential, and I am looking forward to facilitating Onfido’s role within it.
Presidential Executive Order on Safe, Secure & Trustworthy AI: In advance of the AI Safety Summit in the UK Nov 1-2 , the US published its strategy for dealing with the perceived existential threats from the most powerful foundation models. Included were very welcome commitments to introduce international interoperable standards and regulations (via NIST), measures to protect against AI-enabled fraud and deception, civil rights provisions for eradicating bias, and the protection of user privacy (including plans for new federal data protection legislation).
The AI Safety Summit and the Bletchley Declaration: The long awaited UK-hosted AI Safety Summit took place at Bletchley Park, the home of the Allied codebreaking effort in the Second World War, on November 1-2. The summit has focused on the national security risks posed by AI. Twenty-eight nations signed up to the Bletchley Declaration, committing to an internationally inclusive network of scientific research on frontier AI safety, and the use of multilateral and bilateral collaboration to ensure both frontier AI safety and harnessing the benefits of technology responsibly for good and for all. South Korea and France will host the next two summits.
Onfido was pleased to have been involved in both preparatory events in the lead up to the summit, and for invitations to join Prime Minister Rishi Sunak’s business connect with Elon Musk.
AI Act developments: Negotiations for the EU AI Act have entered their final stage with outstanding questions on a tiered approach to requirements, with potential criteria for identifying high-impact models factored from computing power, training data volume, and the financial resources of the provider. There remain nine technical meetings between now and December 6 when the final trilogue will take place.