Engaging with global leaders on privacy, safety and AI

By Julie Brill

Over the past three weeks, I’ve traveled around the globe to listen, learn and engage. Regulators, policy makers, academics, civil society leaders and industry peers have graciously given me their time to discuss a variety of issues related to data protection, #safety , #security , and #privacy in this new era of #AI . It’s been a fascinating trip with insightful conversations, and I’m grateful to everyone who has engaged us.

I’d like to express deep thanks to the Australian eSafety Commissioner Julie Inman - Grant the Australian Communications and Media Authority (ACMA) , including Chair Nerida O'Loughlin PSM , Deputy Chair Creina Chapman GAICD and General Manager Cathy Rainsford ; the ACCC Australian Competition and Consumer Commission, including Chair Gina Cass-Gottlieb , Commissioner Lisa Carber and Deputy Chair Catriona Rowe; the Office of the Australian Information Commissioner , and its Deputy Commissioner Elizabeth Hampton; and 澳大利亚悉尼科技大学 professors Ed Santow and Nicholas Davis .

Similarly, deep thanks to those I met in #Venice at the Privacy Symposium, including Sebastian Ziegler, EDPB Chair Andrea Jelinek , German Federal Data Protection Commissioner Ulrich Kelber, Canada Privacy Commissioner Philippe Dufresne , and Italian Garante Commissioner Guido Scorza . After a brief stop at home to do my laundry, I continued on to #California for #RSAC2023, where I participated in #BrusselstotheBay events and met with European Commission officials, including Gerard de Graaf and Lorena Boix Alonso , as well as 美国国务院 ’s Liesyl Franz and? 美国加州大学伯克利分校 ’s Akasemi Newsome, PhD , among many others.

In all these discussions, from #Sydney to #Italy to #San Franscisco, there were clear currents of both optimism and anxiety. As Thomas Friedman of the 纽约时报 has said, society is entering the age of AI, which has become our generation’s Prometheus moment.

Those I spoke with understood that we are experiencing a major shift in real-time. This era will present unparalleled new opportunities across every aspect of society, from solving major societal issues like climate change, cybersecurity, and advances in health care, to reimagining how we work, learn, and create. Rapidly evolving generative AI applications are democratizing powerful AI technology, empowering all of us. In my travels over the past three weeks, I heard from an economist who believes that AI will create “enormous consumer surplus,” from a healthcare professional that AI will “transform medicine as we know it,” and from a technologist that the opportunity to apply large language models to new innovations are “endless.” After explaining how AI will be infused into productivity tools like PowerPoint and Word and help take much of the drudgery out of knowledge work, the regulators I spoke with said “this changes everything”, “we need to rethink our approach”, and these technologies are “a game-changer.”?

And while there is some agreement on the opportunity, there was also a healthy discussion about the challenges these new technologies will create. Everyone understands the pace of innovation will require all of us to move faster than ever. Regulators and policy makers are rightfully evaluating their role in learning more about these technologies so they can meet their mandates – and meet them quickly – to ensure these systems are safe. Along the way, I found government officials are interested in engaging in dialogue about these new tools. Every stakeholder I spoke with had a genuine curiosity and desire to learn.?

And I learned a lot, too.

No ONE will have all the answers.

No one agency, no one person and no single organization will have all the answers to the complex set of issues this new era presents. As Julie Inman - Grant so aptly articulated, we will need “deep and open discussions about where the challenges and limitations lie and where we can all change and improve.” The solutions we seek, to protect our fundamental rights and ensure the safety of the systems and applications we build, will require input from a diverse set of stakeholders, whose voices are more important than ever. As regulators move forward on this next generation of AI regulation, they will need support from technologists and industry. An example of how we can work together comes from the U.S. Department of Commerce National Telecommunications and Information Administration (NTIA) which launched a request for comment in April to help inform the Biden Administration’s approach to AI-related risks and opportunities. This open process is a great model. The more we can engage together to help governments deeply understand the technology and identify the risks and opportunities, the more we can make progress together.

We cannot let our small differences divide us.

There is a strong desire for like-minded democracies of the world to come together to develop harmonized approaches. Nearly every stakeholder I spoke with noted that promising transatlantic and global discussions are underway. A few key examples: the U.S.-EU Trade and Technology Council (TTC) commitment and collaboration on responsible AI advancements; the OECD - OCDE Artificial Intelligence (AI) Principles to promote trustworthy AI; and the work of the G7 Digital Ministerial on regulatory cooperation and commitments on Data Free Flow with Trust. As these and other global conversations proceed, we should focus on how to create meaningful guardrails that enable innovation while addressing the high-risk uses of these important new technologies.

Consider an iterative approach based on testing and standards.

The technology industry should learn from other industries that for decades have continued to innovate, while protecting people and offering tremendous value to society as they’ve embraced regulation. Take the auto industry, for example. Cars can be dangerous, and to address these risks we created speed signs and safety belts; agencies have created standards that manufacturers must meet; and companies extensively test new models of automobiles to ensure they meet these standards.

Our approach to deploying AI solutions could be similar: regulators could create standards for governance and performance, and companies could test their solutions against these standards.

Whatever approach we take, we know we will not nail it in our first version of these guardrails, and our earliest regulatory approaches will need to be improved as we learn more. But we need to move forward quickly as the opportunity for humanity is simply too great to miss.

We should all continue to listen and learn. At Microsoft we are ready to lean in, and we will continue to share our learnings and welcome feedback along the way. We look forward to being on this journey with you.?