Vinh Nguyen of SailPoint is a former National Security Agency, NSA, chief responsible for artificial intelligence, AI and serves as SailPoint's responsible AI officer. Nguyen joins Rebecca Knight of SiliconANGLE Media and Krista Case of theCUBE Research on theCUBE to unpack how organizations should approach frontier AI. They draw on National Security Agency experience and policy work to examine agent behavior, identity binding for attribution, threat prioritization, governance trade-offs and the practical limits of controls. The discussion highlights implications for enterprise security and cyber resilience.
Nguyen argues that enterprises should avoid paralyzing fear and instead prioritize threats, vulnerabilities and consequences to allocate resources effectively. They emphasize that binding AI agents to identity systems is essential for reliable attribution, that simplistic kill switches introduce new risks, and that organizations should select a few high-impact mitigations, invest in governance and require transparency from frontier companies. The conversation addresses identity security, accountability, attribution, governance and operational resilience, offering concrete steps for security teams and executives. Watch the full conversation on theCUBE for deeper insights on AI agents, accountability and enterprise resilience.
Almost there!
We just sent you a verification email. Please verify your account to gain access to
SailPoint Navigate 2026. If you don’t think you received an email check your
spam folder.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open the link to automatically sign into the site.
Register for Sailpoint Navigate 2026
Please fill out the information below. You will receive an email with a verification link confirming your registration. Click the link to automatically sign into the site.
You’re almost there!
We just sent you a verification email. Please click the verification button in the email. Once your email address is verified, you will have full access to all event content for Sailpoint Navigate 2026.
I want my badge and interests to be visible to all attendees.
Checking this box will display your presense on the attendees list, view your profile and allow other attendees to contact you via 1-1 chat. Read the Privacy Policy. At any time, you can choose to disable this preference.
Select your Interests!
add
Upload your photo
Uploading..
OR
Connect via Twitter
Connect via Linkedin
EDIT PASSWORD
Share
Almost there!
We just sent you a verification email. Please verify your account to gain access to
SailPoint Navigate 2026. If you don’t think you received an email check your
spam folder.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open the link to automatically sign into the site.
Sign in to gain access to SailPoint Navigate 2026
Please sign in with LinkedIn to continue to SailPoint Navigate 2026. Signing in with LinkedIn ensures a professional environment.
Are you sure you want to remove access rights for this user?
Details
Manage Access
email address
Community Invitation
Vinh Nguyen, SailPoint
Vinh Nguyen of SailPoint is a former National Security Agency, NSA, chief responsible for artificial intelligence, AI and serves as SailPoint's responsible AI officer. Nguyen joins Rebecca Knight of SiliconANGLE Media and Krista Case of theCUBE Research on theCUBE to unpack how organizations should approach frontier AI. They draw on National Security Agency experience and policy work to examine agent behavior, identity binding for attribution, threat prioritization, governance trade-offs and the practical limits of controls. The discussion highlights implications for enterprise security and cyber resilience.
Nguyen argues that enterprises should avoid paralyzing fear and instead prioritize threats, vulnerabilities and consequences to allocate resources effectively. They emphasize that binding AI agents to identity systems is essential for reliable attribution, that simplistic kill switches introduce new risks, and that organizations should select a few high-impact mitigations, invest in governance and require transparency from frontier companies. The conversation addresses identity security, accountability, attribution, governance and operational resilience, offering concrete steps for security teams and executives. Watch the full conversation on theCUBE for deeper insights on AI agents, accountability and enterprise resilience.
In this interview from Navigate 2026, Vinh Nguyen, former chief responsible AI officer for the National Security Agency and senior fellow at the Council on Foreign Relations, joins theCUBE's Krista Case and Rebecca Knight to discuss how enterprises can govern autonomous AI agents without succumbing to fear-driven paralysis. Nguyen debunks the myth that only a handful of frontier AI companies understand the technology, arguing that enterprises have the knowledge to shape its development. He explains why security leaders must separate threats from vulnerabiliti...Read more
exploreKeep Exploring
What common misconception about AI would you like to debunk?add
How should organizations think about and prioritize threats, vulnerabilities, and risks—especially regarding AI—when making security decisions?add
How should we handle anonymity, attribution, identity, and accountability for AI agents deployed in enterprises and the wider ecosystem — should they be treated like humans with privacy rights, or bound to identities and oversight?add
Is the AI industry anthropomorphizing AI agents by using human terms like "reasoning" and "identity," and should we be more careful and precise with the language we use?add
>> Good afternoon, everyone, and welcome back to theCUBE's live coverage of SailPoint Navigate 2026 here in Austin. I'm your host, Rebecca Knight, sitting alongside Krista Case, the principal research analyst at theCUBE. So, Krista, are you sleeping okay lately? And I ask because there's a lot of really scary headlines.
Krista Case
>> There's a lot.
Rebecca Knight
>> Yeah, about AI.
Krista Case
>> There's a lot of really scary headlines. And Rebecca, I'm super excited about our next guest because he has deep ties to the intelligence community. And I think there's a lot we can learn as these AI agents start to potentially wreak havoc in our environment. So potentially, yes,
Rebecca Knight
>> right?Anyway, with that, I would like to welcome our next guest. He is Vinh Nguyen, former NSA Chief Responsible AI Officer and Senior Fellow at the Council on Foreign Relations. Thank you so much, Vinh, for coming on the
Vinh Nguyen
>> show.Yeah, thank you for having me. It's great to see you guys, and I love to have a wonderful conversation. Hopefully we can make it positive and optimistic. Okay, not too much, you know,
Rebecca Knight
>> pessimism.Okay, okay, good. Well, I'm glad to hear that because it is an honor to have you on the show. And you've seen AI from the inside of the NSA and now, of course, at the frontier. as in Frontier Labs. I'm going to start with a broad question for the CEOs out there and the board members and even maybe the laypeople who are fearing for the fate of humanity based on these aforementioned scary headlines. Is there a particular myth about AI that is perpetuated that you would like to debunk here and now?
Vinh Nguyen
>> Yeah, I don't think we should fear it. It's a technology, and I believe in humanity that we can harness, learn, master, control the technologies. I think that there are legitimate concerns and fears, but my take is that we shouldn't fear it. We can learn how to adapt it. I think the one myth is that we really assume that frontier AI companies know everything, and that's a myth. I think we should be empowered that we can also make the contribution to frontier AI development and deployment as well. It doesn't have to be only a few companies or a few people who can understand this. I think we have the capacity, the knowledge, and experiences to guide this technology forward.
Krista Case
>> And in terms of guiding it forward, Vinh, I'm curious, based on your experience in the intelligence community, spending a lot of time thinking about how to operate against these adversaries that are very adaptable, but they potentially hide their actions and they find gaps between systems to exploit. What are the top lessons from this experience in the intelligence community that you would maybe advise to an enterprise that is trying to position against agents that are doing these very same things?
Vinh Nguyen
>> Yeah, in national security, we have to think about the threats and then prioritize the threats. And I think a lot of people are mixing the different concepts and it makes it really difficult to make decisions. I would say for national security, you always have threats. You have China, Russia, Iran. There are terrorists and cartels, but you have cyber threats. They are threats, but the threats can only manifest if you are vulnerable. Then if you're vulnerable, then what are the consequences that you are going to deal with? Then you can make a decision on prioritizations of what threats you're dealing with. It's just when we talk about AI and enterprises and they think about risk, they go directly to the end and don't think about what are the threats to the business, what are the threats to just the company's posture overall, data, you name it. And so they have to rethink rationally and logically that you have a lot of risk. We always have risk, but what are the threats, the vulnerabilities that you are dealing with, the consequences, and then prioritize a few things to actually minimize and reduce the risk. Otherwise, we can have infinite money and capital and won't be able to reduce our risk. I think for enterprises, they have to think adaptively and be able to make better judgments and prioritize and actually do. A lot of companies aren't thinking it through, they're not prioritizing, they're not putting in the practices and the talents or the investment to actually do it. So I highly recommend you find a few things to do, prioritize them and just do them.
Rebecca Knight
>> Because what are the unintended consequences of, say, slowing down out of fear?
Vinh Nguyen
>> Well, I think depending on your industry, by slowing down, you think that you have the control, but you actually don't, all right? And so you have unintended impact. So when you're telling your companies and your workforce that, yeah, we won't be able to like we don't allow you to use this, or you shall not be allowed to run this or that, people will find ways to actually do things that you prevent them to do. It's just people. And so suddenly you open up a new security problem that you are not ready to solve. And so what you really have to do is you have to listen to the people and meet where people are at and be able to adapt and provide the tools, even if it's not perfect, but at least you can channel them and then provide a level of governance and guidance so you don't end up fighting like 5 different problems that you don't have time for.
Krista Case
>> And Vinh, you talked about humans taking actions that we maybe didn't anticipate. Naturally, we're seeing AI agents are 100% doing the same. We're giving them an objective and they're finding a way to accomplish it. So based on your work around cyber attribution and what you've seen on the human side, when something maybe goes wrong or has an unintended consequence based on the actions that an AI agent took, how do you think about establishing the chain of responsibility and who's accountable at the end of the day?
Vinh Nguyen
>> Yeah, we have to actually take us a while back to when we built the internet. We really value the anonymity of the internet, right? Openness and freedom of expression and being able to communicate. No one knows you're a dog. No one knows you're a dog, right? Exactly. That was a long, long, long time ago. And the problem is that because it's anonymous for a long time, adversaries start to misuse it and use it in order to attack and exploit different systems. We learned that lesson already. Let's not relearn that because we already know it? And so as we unleash AI agents into the enterprise and into our partners ecosystem, we have to ensure that we can start right away with attribute, we'll be able to make attribution and be able to really bind AI agents to an identity system or something that we can control. Because if you don't do that right from the beginning, I really like the experts around here for this conference, is that they really think deeply about it. But we actually continue to assume that these are still like humans, but they're not. We have to rethink and say, by the way, they should not have privacy rights. They should not be able to hide their communication. So, we have to think it's like, okay, we treat people, we should not apply how we treat people and apply it to agents because we want to protect the right of people as a democracy. But for agents, we should not apply the same
Rebecca Knight
>> principle.It's not a democracy. It's not a democracy for agents. It's not a democracy for conscious entities.
Vinh Nguyen
>> Exactly. And so, why are we like So, we have to think carefully because we use the same language, we use agents, but everything should be accountable and all agents' actions should be accountable to an individual, an enterprise, a corporation that we have the laws to bind. That is our contract. In a democracy, we have a social contract. This is our accountability, openness, and recourse. That's like so core to democracy. But we should not somehow build agents and say, you know what, agents can do whatever and don't have to abide by our democratic principles. Well, what technologies are we building then?
Rebecca Knight
>> Well, do you think you're an anomaly in the industry for saying these things? Because I do feel like there is this real anthropomorphism, anthropomorphization of AI agents that's also coming from the industry itself.
Vinh Nguyen
>> Well, because we lack creativity to find new words.
Rebecca Knight
>> Yeah, you're right.
Vinh Nguyen
>> It's a lack of imagination. So, for example, when an advanced AI model does logical chaining of context and content, we call it reasoning.
Rebecca Knight
>> Right.
Vinh Nguyen
>> Because we don't have any other term to explain. Because we're like, oh yeah, we reason, but we don't say it feels. Yet.
Rebecca Knight
>> But that's what Claude is telling me it's doing. I'm cogitating, I'm ruminating, I'm reasoning, and so it makes me buy into it. But you're right.
Vinh Nguyen
>> Yeah, yeah, I think we just have to be careful of the language that we use. Even the word identitymeans different things to different people. If you're in Washington, D.C., identitieswhen people talk about it, it's more like political identitiespeople's identities. They're not thinking about agents, and then when you have agents, they're like, "Huh?" Right? I think we can find different ways to talk through it, but sometimes we get lazy and we're like, "You know, it's close enough," and then we just slap it on. Sometimes we really have to be really careful and say, "You know what? It may not be reasoning. It is doing something else. It is not hiding. It's doing something else." We may want to use different words, and we can scientifically explain and define it better.
Krista Case
>> And Vinh, when we think about accountability, we've seen Glasswing, Daybreak, these kind of industry consortiums. What's the role that you think that these consortiums play in kind of helping to maybe establish some of this accountability and maybe help to progress our mentalities around this?
Vinh Nguyen
>> Yeah, I think this is a bigger context and view than we think. When you really think about the international order, it's really falling apart, and we have to rebuild the order into the future. We're just in the nascent stage of what the new world will look like. And so in this environment, we are actually renegotiating the contracts between governments, industry, people. And so when the government is not stepping up to do a certain task because it's testing the social contract, then the responsible parties or the people who feel like maybe it's their responsibility to step up and solve the problem get established. You can see various efforts like Glasswing, Daybreak, it's really an industry attempt to say, okay, if there's no backstop from the government, What are we going to do? It is really a manifestation of new leadership and mechanisms in the new world. How successful it is? I'm not sure. The incentives are not really aligned. We design governments to do very different things, like for public goods. Companies are not built for just public goods. They do some public goods, but I think this is a negotiation. I think it's a good signal that something is really wrong and we need to do something about it very quickly. And then I applaud the frontier companies for really stepping up. And they didn't have to do it, but they stepped up to do it, be it Anthropic, OpenAI, Google DeepMind, and the like.
Rebecca Knight
>> And the question is, how much trust do we put in those companies to do that?
Vinh Nguyen
>> And they have to earn our trust. And I think we should, we as the people, should expect the companies to work with us to build our confidence in them. I think in this world we don't take anything for granted and we should not trust anyone. I think they have to earn our trust, meaning they have to be responsible, they have to make sure that our data is private and not leaked or used in an irresponsible way that is, you know, kind of extract who we are, right, as a people. And yeah, they just need to do things that we would like, and we have to speak up when we like it and speak up when we don't like it. It's a negotiation right now, and people somehow find it surprising, but to me it's an opportunity that we can speak up to demand what we want in the future.
Krista Case
>> And Vinh, another important negotiation as part of these conversations that we're having as an industry is the controllability over the AI agents at the end of the day. I've heard you have an interesting take on kill switches. Are they sufficient? Are they not? Can you walk us through your take on that and I guess what we should be looking for when it comes to really having control over these agents, especially when they start to take action on our behalf.
Vinh Nguyen
>> Yeah, I love these kill switches. I think they come up every 10 years. Every 10 years there's a kill switch.
Rebecca Knight
>> Oh, God, you're done.
Krista Case
>> Right?
Vinh Nguyen
>> It's important because I really worry about this. Yeah, I think in the early days, if you recall, the internet kill switches, when we're like, oh, when things go wrong, we hit the kill switch so that no attackers can attack us. And I was like, well, you just kill-switched yourself? And so my problem with kill switches are really threefold. One is that when you introduce a new capability like a kill switch, you also introduce insecurity into the system. And so it's like you secure, your home, your fences, and then suddenly you open this door and you're like, please look here and then break through this so you can burn the house down, right? And so the insecurity is really a concern. So you have to secure this kill switch so that your adversaries don't use the kill switch against you. The second is that in frontier AI capabilities, AIs, in terms of running these, are spread across multiple data centers. So when you kill switch a data center, did it really take away the AI or not? And lastly, the practical thing is, when are you going to hit the kill switch? You have it, but you have to know when. And we don't know when yet. The kill switch is kind of useless. I think that's what the community is trying to build is, to be able to observe, be able to trace AI, be able to understand the behaviors, and when things go wrong, intervene. We don't have that yet. If you don't have the mechanism to decide when to hit the kill switch, then the kill switch is pretty, but it's not going to do much for you.
Rebecca Knight
>> Okay, we need to end this on a positive note here because you have a lot of tech optimism, but there also is a lot of trepidation I think that a lot of people approach this with. So you're here at Navigate. I'm interested, is there something that surprised you about how the community, the enterprise security community is approaching this agentic era? Something that gives you some hope about the way industry is?
Vinh Nguyen
>> I always believe in people. I think we are creative, we are resilient, and we are adaptive, and we are willing to do hard things and solvereally difficult problems. So I believe in humanity. I believe in us. I think we will overcome this. And you can see the people here, years ago there are few people working in identity security. Now you have like 1,000, there's a lot of people. People do step up and people do care and people want to build a better world. And if you believe in that, we will be able to overcome the challenges that we face. So I really believe in people. I believe in community. And today there's a lot of fear and uncertainties, but I think at the end of the day, we have to trust and empower ourselves because we are more powerful than AI. We can do it and we should be confident that we can.
Rebecca Knight
>> I want you at the table with the White House next time. Vinh, I'm going to get you an invite. You got to be there. Vinh, thank you so much. A real pleasure having you on the show. Really great conversation.
Vinh Nguyen
>> Yeah, thank you.
Krista Case
>> Thank you, Vinh.
Rebecca Knight
>> I'm Rebecca Knight with Krista Case. Stay tuned for more of theCUBE's live coverage of SailPoint Navigate. You're watching theCUBE, the leader in enterprise tech news and analysis.