Intersection of Technology and National Security with Deputy National Security Advisor Anne Neuberger

THE WASHINGTON FOREIGN PRESS CENTER, WASHINGTON, D.C. 

MODERATOR:  Good morning.  Good morning, everyone.  Welcome to the Washington Foreign Press Center briefing.  My name is Zina Wolfington, and I’m the moderator today.  It is my pleasure to welcome our distinguished briefer, Deputy National Security Advisor for Cyber and Emerging Technology at the National Security Council Anne Neuberger to discuss an intersection of technology and national security.  This briefing will end at 10:35 a.m.  It is on the record, and we will share the transcript of the briefing with you shortly after the briefing.  After we hear from Anne Neuberger, we will begin a question-and-answer session.  And with that, I’m going to turn this program over to Ms. Neuberger.   

MS NEUBERGER:  Thank you so much.  Good morning everyone.  It was so interesting to look at all of your bios and see just the wealth of knowledge, policy experience, and backgrounds in the room.  So I’m particularly looking forward to the question and answer session.  I’ll just make some remarks at the top because I know you’ve had a really excellent program.  You’ve been traveling, and you’ve heard a lot.  

So technology has long shaped foreign policy.  Countries which adapted technology growth power their economies, attracted skilled labor, drove productivity and economic growth.  And we certainly know militaries that adapted technology were able to be forces on the global stage – from advanced intelligence, precision intelligence, to robotics for hazardous missions to keep soldiers safe.   

So technology fundamentally has shaped geopolitics and economics for a long time.  And we can see the advancements in technology that are poised to define the geopolitical era of the future; for example, the combination of AI, advanced telecommunications, and sensors will generate leap-ahead breakthroughs in drug discovery, food security in an age of extreme weather, and clean energy in era where we’re optimally fighting climate change.  It will also enable novel military and intelligence capabilities that will shape our collective security.  And this is a group that has covered technology and policy for a long time, so I know you see that arc both with its promise and both with its peril.   

So in the United States, we’re carefully considering the national security implications of AI, including risks and opportunities, as well as tangible trust and safety mechanisms that could help us achieve the promise of AI that can help us have our confidence – the confidence of our citizens in AI’s use in our economies and in our society.  And we want to achieve that promise together with key allies and partners, which is why you are here.  Because international collaborations can ensure we all have equitable access to the promise of emerging technologies.   

Recently, I met and convened a group with the ambassador – German’s ambassador – Germany’s ambassador to the U.S. – we convened a group of female ambassadors in D.C.  We said often it’s – for women it’s harder with less of a seat at the table in technology and technology policy.  Let’s convene the female ambassadors in D.C. to share knowledge, to add knowledge, and to equip them to be effective policymakers in this space.  And two of the women at the table said something – this is now several months ago – which has stayed with me.  Two of the female ambassadors – one from Africa, one from Latin America – said, “You know, in the past, our societies have been sometimes left behind or a step behind with technology.  Keep us at the table for artificial intelligence.  From data, for compute, for algorithms, give us an opportunity to have equitable access to this technology.”  And it’s really something we keep in mind and are very focused on; of course, both in managing the risk but also in achieving the promise.   

So a first step in that way, we’re focused on joint work with experts from other countries.  Last year between the U.S. and the European Union we signed an administrative agreement focused on AI for public good, to drive both progress in AI and related privacy protecting technologies in five areas: one is health – there are 11 areas of partnership underneath, including building advanced models for more effective cancer detection, building advanced models for more effective cardiac treatments.  There’s a second line of work around extreme weather prediction.  Can we predict – in the last six months alone, we’ve seen so much hardship based on extreme weather: flooding, fires.  Can we better predict those to enable both people – emergency management but also agricultural optimization?  If there’s going to be higher temperatures, do we plant differently?  Do we irrigate differently?  Do we plant in different parts of the world as well?  Electric grid optimization because our energy grids are big parts of carbon consumption and generation.  I mentioned already extreme weather and climate forecasting.   

There are a number of partner projects underway.  I was just in Brussels last week to convene a group to discuss progress, and our goal is – after the U.S. and EU make good progress bringing that to a broader set of countries – to bring in country-specific data, country-specific – and build models and make those advancements together.   

I want to mention one specific area in the area of climate, because that’s something I know of concern in many of the countries here.  The European Union has an Earth system digital twin model called Destination Earth.  And we have similar models as well as detailed climate data in the U.S. that will be powerful to bring together to generate AI algorithms to predict the impact of climate change on agriculture – and as I mentioned, in order to generate both new seeds, new irrigation policies, new planting policies.  So our goal is to work together to tackle some of the world’s toughest challenges.  And certainly we see with UNGA now, right, when we look at the – certainly the deliverable AI for the public good. 

Our international cooperation is focused on managing the risks and proving that AI can be done in a way that respects human rights and fundamental freedom, while providing that benefit.  We believe we can generate the benefit of better cancer prediction models without also predicting individuals’ private health information.  And that’s one of the goals as well.   

I want to talk a bit about other partnerships with some of the countries here in the room, and then I’ll break for questions. 

Partnership with India – we’re working together on joint collaboration on trustworthy and responsible AI, advancing AI standards and measurement, AI education in the workforce.  Certainly that’s an area of concern in all of our countries.  There will be new jobs; how do we handle the transition and the training from jobs that may be impacted?  And the United States supports India’s leadership as chair of the Global Partnership on AI.   

Partnership with South Korea – President Biden and South Korean President Yoon have decided to deepen and broaden cooperation on critical and emerging technologies like AI.  We also signed a far deeper cyber security partnership during President Yoon’s recent visit.  And we’re working together specifically with Prime Minister Kishida of Japan to advance efforts to develop standards and ensure trustworthy safe and secure artificial intelligence.  I would note on the area of emerging technologies, we launched a trilateral partnership between Japan and South Korea focused on cryptocurrency, and detecting illicit use, specifically North Korean illicit use of crypto that is driving advancements in their missile program.   

And I’m happy to talk as well about the specific steps the administration has taken.  As you know, in September, just yesterday, the White House announced an additional set of companies that have committed to the voluntary commitments.  In August the White House launched the AI Cyber Challenge to get an advance for defenders.  We know there will be folks on defense, and will be folks on offense.  We want the defenders to be a step ahead.  In July, the President convened seven leading companies to get the first set of voluntary commitments, and certainly over the spring the President met with top AI experts and researchers, convened consumer protection, labor, and civil rights leaders to discuss risks related to that area, and release a national AI R&D strategic plan, an AI Bill of Rights, last October – almost a year ago. 

So certainly a lot of action here.  I know you have questions you want to talk about, so I’ll open it up to your questions, and so exciting to be here with you today. 

MODERATOR:  Thank you for this remarks.  Now we would like to open for Q&A session.  Please raise your hand if you have a question.  If I call on you, say your name and your country.  And we’ll go with Felix. 

QUESTION:  I have two connected questions.  I mean, while there’s obviously enormous positive potential stemming from AI, is – especially generative AI will also help malicious actors, both state and private actors, to do disinformation campaigns, influence operations more effectively.  What are you seeing and what are you expecting to see in the upcoming elections, and how is the administration preparing for that?  And related to that, what are you already seeing from state actors, from Russia and China, abroad? 

MS NEUBERGER:  Okay.  So Felix, you asked a lot of questions there.  Let me try to get at each of them.  So first, a number of the voluntary commitments related to protections we believe are very much needed in this space.  Those were corporate voluntary commitments.  The White House is also working on an executive order.  And as you know, and I believe you were in Congress yesterday, our Congress is working on potential new laws, because the executive order is what is the full boundary of what can be done under current laws. 

So first, within the voluntary commitments, you saw for example companies committing to add a watermark, a mark on video or image data.  We believe that is particularly needed in the area of countering disinformation so that we can educate our populations to say look for a watermark, or potentially one that is both there and one that is hidden to make it harder to remove, and know that that is AI-generated content.  And we can educate our populations to say if it’s AI-generated content, ask, is it real?  As an example, just several months ago – and then we must have the rapid response (inaudible) for that.  You may recall several months ago somebody tweeted a picture of the Pentagon on fire.  The local fire department very quickly responded, said this is not the case, this is not true, and had a response protocol.  We need to also create those response protocols across our state, federal, and local governments so that they are prepared to respond. 

So companies – so that’s how it is layered.  Companies placing a watermark by default give us a way to technically determine is this real or not, is it AI-generated; give a way for consumers to be aware as well as a way to drive more rapid response.  I think it’s certainly something we’re watching in the upcoming elections.  Disinformation existed before generative AI.  Generative AI we believe – we’re concerned – do make it easier to generate better deep fakes, more focused disinformation, and as a group of governments we’re working together and with the key companies to think hard how to address. 

MODERATOR:  Thank you.  Bilal. 

QUESTION:  I’m Bilal Hussain, from Pakistan.  Much of the tech companies are American, which we use, like Facebook and such companies.  So the legislation in America White House is looking for in AI will have – will it try to protect countries like Pakistan with has – which the government doesn’t have (inaudible) or international big companies like – most of the companies that have American origin?  So small countries will be – will the White House be giving some thought of (inaudible) from its own company, not the Chinese (inaudible) but its own company?  Will we be having some protection in the AI or in these sectors? 

AMBASSADOR NEUBERGER:  It’s a really good question.  So first, we’re approaching this not only as the U.S. but also on the international stage.  As you know, there’s an effort in the G7, there is an effort under the Hiroshima process, to ensure that as a group of countries we’re setting international norms as countries.  And then our goal is, both in the executive order that’s focused, as you know, on the U.S. but also on the potential legislation, that will guide the way the companies operate around the world.  And that is our goal.   

By setting the standard in law, we also are working with other countries to say this is what we believe are the appropriate controls so that can then be used by other countries to enforce as well, but also as a way for us to say how do we balance innovation and risk.  And you saw when you were on – in the Hill yesterday how much folks on the Hill are thinking hard about these issues, bringing people in from civil society, from academia, and the countries involving others to really outline the way ahead that isn’t just for the U.S., but that sets the international norms, sets the – what we believe should be the norms for behavior in this space as well. 

MODERATOR:  Soonmin, do you have a question? 

QUESTION:  Yes.  I am Soonmin Hwang from South Korea.  Thank you for your time today.  It seems that AI technology will play a decisive role in national security like nuclear weapons.  I’m wondering what is the United States response strategy to military AI (inaudible) in countries such as China, Russia, North Korea.  And what specifically is discussed cooperation between Korea and United States Government in this area? 

AMBASSADOR NEUBERGER:  It’s a very good question.  We are certainly very focused on how to ensure that military technologies are – that we consider how AI combined with military technologies can bring additional risk.  We talked about earlier, for example, in the area of more precise intelligence collection, certainly we’re concerned about authoritarian governments using AI to maintain close surveillance on citizens.  One can imagine, right?  Because of the advances in computer vision, countries, authoritarian governments being able to bring all of those images together to generate precise data about citizens in a way that impacts individuals’ freedoms, and that is certainly something that the United States seeks to outline what should be appropriate protection.  The AI Bill of Rights talked about that piece as well.  What should citizens expect from their governments in terms of those protections? 

We are working internationally.  I mentioned the G7 and the Hiroshima process.  I mentioned the upcoming UN, working closely supporting the secretary-general’s work in this space, the convening he will be doing in this space to ensure that we work with international partners to put in place international norms to govern use of AI so that it could be a force for good and that we manage the risks, some of which you outlined.   

MODERATOR:  Sashana. 

QUESTION:  (Inaudible) — 

AMBASSADOR NEUBERGER:  Talk a little louder. 

QUESTION:  Oh, sorry.  I’m saying that we’ve heard about the blueprint for AI as well as voluntary commitments from companies, and now that government trying to get bipartisan support.  But I want to know how difficult do you believe getting bipartisan support for an overarching (inaudible) regulation will be with America’s current political climate. 

AMBASSADOR NEUBERGER:  From what we have observed, the Hill is really seized with the importance of AI for our economy, for our society, for our national security.  Senator Schumer is convening bipartisan processes to drive rapid progress.  So we feel confident that the Hill is taking these issues on in a bipartisan, thoughtful way and working quickly to think through the complexities of driving innovation, managing the risks, considering how we train a workforce for an era of AI, considering how we take advantage of AI in manufacturing while also ensuring that workers are protected, and thinking through how individuals across the society from different backgrounds, different experiences, different educations can be a part of a future AI economy.   

MODERATOR:  Hugo. 

QUESTION:  Thanks.  I’m Hugo Seneca.  I work for Expresso newspaper from Portugal.  Would it be wise or would it make sense to create a kind of a Geneva convention in United Nations for AI?   

AMBASSADOR NEUBERGER:  So we know the UN secretary-general is very focused on convening a group of international experts to think through an approach for the future of AI.  He’s thinking very thoughtfully about his SDGs, the societal development goals, how AI can help.  Some of the areas I mentioned in our AI for Public Good, like extreme weather prediction or agricultural optimization, very much align with the secretary-general’s thoughts.  And we met recently his ambassador, Amandeep Gill, was here.  We met with Ambassador Gill to talk about the work the secretary-general has asked him to do at the UN.  So we very much are supporting the secretary-general who take steps forward.   

And we believe that the work the U.S. has done in bringing together companies to define and make voluntary commitments – whether that’s in the area of watermarking, whether that’s in the area of red teaming models before they’re released, whether that’s in the area of greater transparency around what data models are trained on – all of that is foundational to getting citizens’ trust in use of AI and in considering, as well, countries’ commitments around use of AI. 

MODERATOR:  Ruslan.      

QUESTION:  Ruslan from Kazakhstan.  Do you have any specific plans on the operation with Central Asian countries, with those who wish, like, Russia and China have a little bit of a more – not impact, but that they are more prone to Russia and China’s AI (inaudible)?   

MS NEUBERGER:  President Biden really sees technology as a key part of the way we engage countries around the world.  We’ve deepened, for example, I believe our cyber security partnership with Kazakhstan.  You’ve seen various technology agreements that have been signed around semiconductor, emerging technology partnerships – for example with Vietnam, President Biden’s recent visit there – certainly a part of our work with countries around the world.  As we look at Central Asia, there are so many opportunities there to use AI for public benefit – whether it’s extreme weather prediction, agriculture optimization, electric grid optimization, communications, right.  You have large – large countries, distributed populations.  How do we ensure that we can use AI to determine, from a communications perspective, access in a broader way.  So it’s very much a part of our goals in our foreign policy.   

The first part of the efforts I talked about with regard to delivering on the voluntary commitments, working internationally in the G7 process, is to share our view of what are the safety and controls that need to be in place.  How do we glean the trust of populations that AI can advance their economies and their societies?  And then how do we work together in those areas as well?  So we really see that as sequential in terms of those commitments on trust and safety, gaining the trust of the societies, building deeper partnerships between our countries on data, on compute, and on algorithms as well.   

MODERATOR:  Leo.   

QUESTION:  Leo (inaudible).  I’m curious to know what you’d say to the – to women from Africa and Latin America who said they don’t want to be left behind, because as it is in Zimbabwe, we feel left out.  There are (inaudible) do not have access to right now.  And some of the reasons we are given is they’re trying to balance benefits with drawbacks.  They’re kind of scared of how it will be used by some of our governments, but if you look at the totality of the benefits that AI brings, we are falling behind.  So I’m curious to know what your answer is to that.  And what is the U.S. doing really to get some of these smaller countries, like mine, at the table to access all of this, yes?   

MS NEUBERGER:  So it’s exciting to see some of the AI work that is being done in Africa.  I think there are some leading models in Kenya and believe there are some startups in Nigeria as well.  There’s certainly – it’s exciting to see some of that work.  And I think as we look at the progress that has happened in open-source AI, that allows broader access.  Originally, I think several months ago, the model we were seeing of AI was very large companies training very large amounts of data at great cost.  Over the last few months, we’ve seen how building on that already-trained data, smaller startups are refining and building more focused models for specific things – whether that’s specific companies, like Bloomberg’s specific data for financial purposes – whether that’s specific to a country, like a given country’s medical records for advancements for a given ethnic population or age group.   

So I think the progress in open-source has brought down the cost and really democratized AI.  And certainly seems like the national AI research effort in this country would work to make compute available to researchers – as well as, as I talked about, things like the administrative agreement with the EU, which we intend to expand, which seek to first build models based on countries that have large amounts of data, and then take those and work with countries who may have smaller amounts of data to refine them and have country specific values.  That’s the approach we’re thinking about because we certainly see the point you raise.  There needs to be promise, and there’s an opportunity to bring that promise in a very equitable way.  But it will take focus and work, but I think that’s very much the intent.   

MODERATOR:  And we have time for one more question.  Aziza.   

QUESTION:  Kyrgyzstan.  My question is:  America and Central Asia are extremely, like, on different level of cyber security for now.  And our region, especially my country, is known, like – it has political instability in it now like at least 10, 15 years.  And how do you think – what long-term cyber security technologies and strategies will help to have stability and security in Central Asian region? 

MS NEUBERGER:  So that is a very strategic question; I will do my best to answer it in a couple of minutes.  I must say the English everyone speaks here is so excellent, so congratulations to you.  Cybersecurity is an area where – it’s a transnational challenge for us, right.  You can have individuals sitting in 10 countries, using infrastructure in another 10 countries, conducting cyber attacks on a final set of countries.  So fundamentally we see each country’s protection is within the country, but also working Internationally.   

So for example, at the end of October, October 31st, we’re convening 47 countries here for the third meeting of our International Partnership to Counter Ransomware.  Criminal cyber attacks are disrupting hospitals and schools and companies all around the world.  There’s a disruption – multiple disruptions in the U.S. as well.  And it’s the top thing that, when individuals from other countries come to talk to us, they come to talk to us about.   

So we use this initiative to do capacity-building, regional capacities.  For example, we did a capacity-building initiative of a set of African countries to do capacity building on tracing illicit use of crypto that funds illicit from movement of drugs, to funding criminals around the world.  We do it in that way.  We will be working together on a first diplomatic statement related to this as well.  So we very much are committed to both doing capacity building with countries, particularly with regions so countries can work more closely together, as well as sharing threat information.  Under our Department of Homeland Security, we have an active program where we’ve expanded the number of Computer Emergency Response Teams that we both share with other countries when there’s an emergency, as well as the information we share with Computer Emergency Response Teams around the world.   

And finally, in the cyber security guidance that we publish on the web, which represents our best knowledge across our law enforcement, our intelligence, and our cyber security community that we seek to bring.  I think as you asked the question, I’m thinking to myself that it probably would be a nice idea for us when we publish the cyber security advisories to do so and translate them into a number of languages to make them more available.  So thank you for your question.  I will take that for follow-up.   

MODERATOR:  So with that, unfortunately we don’t have any more time left.  I would like to thank Anne Neuberger for briefing us today, as well as journalists in attendance.  This concludes our briefing.   

MS NEUBERGER:  Thank you.