DSS 2026 / Day 2 / Amber Hall
Security is a leadership problem
Josselin Feist
Source: YouTube automatic caption track · Exported 2026-11-16. Timestamps link to the source video. Cue end times were not supplied.
Transcript
0:01Okay. Hi everyone. Um, welcome to this
0:03presentation, security is a leadership
0:05problem. So, first thing first, who am
0:07I? My name is Jocelyn Feist. I'm a
0:09security researcher and I've been in the
0:11blockchain space for almost nine years
0:12now.
0:14Through the years, I had the chance to
0:15work with many protocol, DeFi, layer
0:17one, layer two. Um, and I have worked
0:20with protocol with various stage of
0:22development, various maturity.
0:24This presentation is a personal take
0:26based on some of the observation I have
0:27made through the years.
0:29Oh.
0:30Sorry.
0:32So, when I started my career as a
0:34security researcher, my main focus was
0:36to find bugs. My kind of thinking was
0:39that if I can find all the bug in the
0:40code base, my job will be basically
0:42done. Security is a technical problem,
0:44so I need to find technical solution.
0:46Over the years, I've realized that many
0:48bugs stem from incorrect design
0:51decision, architecture decision, or
0:54shortfall in the testing practices. So,
0:56I ended up focusing a bit more into
0:58helping protocol and developer to think
1:00about their design and the long-term
1:02security.
1:03Uh, that helped, but over time, I've
1:06realized that this is not enough.
1:08Even if you find improvement you have to
1:10make in the design, even if you find
1:12shortcoming in the testing, someone has
1:14to make a decision. Someone has to
1:16decide, are we going to allocate
1:17resource? Are we going to change the
1:19product? Are we going to change the
1:20timeline? And this made me realize that
1:23security is not just a technical
1:25problem. And this is sometimes a bit
1:27difficult because as engineer, we focus
1:29on technical problem. But at the end of
1:31the day, uh, a lot of the decision are
1:33going to be made by the leader of the
1:34protocol, and as a result, security is
1:36not just a technical problem, it's also
1:38a leadership problem.
1:41We knew it was a weak point, but we
1:44thought it was okay. That quote stuck
1:46with me. I have heard it many time from
1:49different team with little variation
1:51after an incident.
1:53When you were looking back, when when
1:54the team was looking back why an
1:56incident happened, they realized that it
1:58was not something completely obscure
2:00that happened. It was something that
2:01they knew about, but thought it will be
2:03okay. That could be a low severity issue
2:06that was acknowledged and never
2:07revisited. That could be a situation
2:10where the smart contract were actually
2:11robust and and were in a good point, but
2:13the team never considered operational
2:15security and as a result, that's why the
2:17attacker
2:19led to exploit.
2:21We have in the team we we have in the
2:23industry really good and strong
2:25technical team. They care about
2:26security, they spend a lot of time, but
2:28sometime they still fail because at the
2:30end of the day someone has to make a
2:32decision, someone someone has to decide
2:34are we going to go live or are we going
2:36to spend more time looking for issue.
2:39And this is a leadership decision. Now I
2:41know you might have heard similar take
2:43in the past and I know leadership can
2:45sound like a bit of a disconnected
2:48topic, but I so know firsthand that many
2:51protocol leader in the space do not
2:53consider security as their
2:54responsibility.
2:56As a result and as we are going to to
2:57see through this talk, most of the time
3:00issue are never revisited, risk are
3:02never revisited, exposure can scale
3:04faster than the confidence in the
3:06system. There is no security ownership
3:08within the protocol team and more than
3:11often perception of security becomes
3:13more important than the security itself.
3:17So who here has deployed a protocol
3:20within the past let's say 12 months and
3:23could point me to the list of non-issue
3:25on the list of risk accepted and would
3:28be able to tell me when they have been
3:30revisited.
3:31Probably only okay, only a few people,
3:34right?
3:36In reality, you are going to make
3:38trade-off, you have to make decision.
3:41And you know, when you need to go live,
3:42you're going to accept some of the risk.
3:45But in reality, once you accept a risk
3:47even temporarily, most of the time they
3:49become permanent.
3:52And the problem is not necessarily that
3:54you make a trade-off, the problem is
3:55that you never revisit them. However,
3:58your protocol will evolve. You will have
4:00new integration, you will have new
4:01asset, you will have new components. And
4:04if you don' spend more time trying to
4:06think about how to re-evaluate and
4:08re-assess the risk, something that was a
4:11minor risk can quietly become a
4:12permanent risk within your system.
4:16So this is where as a leader you need to
4:20periodically re-evaluate and and your
4:22risk and allocate resources for that.
4:24This is a boring task. No one likes to
4:26do that. And if you do it, you know, you
4:29won't be able to tell your investor that
4:30you have made progress. You won't be
4:32able to generate revenue. You won't be
4:33able to kind of like, you know, show
4:35your user that you have like a new
4:36report to publish. And most of the time
4:39when you are going to re-evaluate your
4:41non-risk, no decision will be made
4:43because the trade-off that you have made
4:44might be still true today.
4:47So this is a bit uncomfortable, but it
4:49is why this is a leadership decision.
4:52Next time you deploy a protocol, you can
4:54tell your engineer, we are going to
4:56deploy next week, but in six months from
4:58now
4:59we are going to re-evaluate your risk.
5:01This implicitly will tell your
5:03engineering team that they need to
5:05document and own the risk.
5:07And this is how as a leader you can
5:08create a culture in within your company
5:11where the management of risk is
5:12important.
5:17I frequently hear protocol leader tell
5:20me that they don't have the budget for
5:22security.
5:23Now security is not just about paying
5:25for audit. It's also about controlling
5:27your exposure while your confidence in
5:30the system is limited.
5:32Most of the protocol team won't have the
5:34budget to do multiple round of audit,
5:36formal verification, fuzzing, and like
5:38all the different techniques. But they
5:40can decide to not expose immature system
5:43to unlimited loss.
5:46We have seen multiple time in the space
5:48small experiments, small idea that
5:50overnight become a high value target.
5:52And this is a problem if your security
5:54budget and your security confidence in
5:56the system does not allow you to
5:58properly assess your system. You do not
6:00want to be in a situation where your
6:02exposure scale faster than your
6:03confidence in the system.
6:05The good news is that as the protocol
6:08leader, you have a lot of lever that you
6:10might never know you have.
6:12For example, when you are going to
6:13create your first version of the
6:14protocol, you can you can think about
6:16reducing the scope of what you are
6:18building. This means we're moving some
6:20integration, we're moving complex um
6:23complex part of the protocol if you
6:25don't need them for V1.
6:27You can directly limit exposure, you can
6:29control your exposure by having TVL cap,
6:31by having rate limits.
6:33You can have centralization point that
6:35are will going to help you to control
6:37that, for example, pausing mechanism and
6:39this type of things. Now, this decision
6:41and this trade-off are a bit
6:43uncomfortable because they're going to
6:44compete with competitiveness, growth,
6:47revenue generation, but this is why they
6:49are leadership decision.
6:51If you don't do it, no one will do it.
6:53And this is where great leader think
6:56ahead and think about what they're going
6:58to build and how their security budget
7:00will allow them to build what they want
7:02to build.
7:04Now, this can become difficult to
7:06evaluate if you don't have any security
7:09expertise and any security ownership
7:11within your company.
7:13If you look at traditional financial
7:15institute, they have a lot of security
7:17expertise. They have a CISO, they have
7:19an internal structure, they have
7:20security DevOps, they have security
7:22engineer.
7:23If you ask most of the DeFi team today
7:26how they handle their security, they are
7:28going to tell you we are done audit.
7:31Most of the industry has externalized
7:34security expertise to third-party
7:36provider.
7:37That model made sense when the industry
7:39was scaling and expertise was scarce,
7:42but it is not a good model today.
7:44Protocol team, blockchain team, and DeFi
7:46team needs internal security and they
7:48need security ownership.
7:50Now, what I have see sometime happen is
7:53that one engineer on the team become the
7:56owner of security, but that engineer has
7:59no lever. They cannot make decision,
8:01they might identify risk, they might
8:02identify concern, but they have no
8:05ability to change any direction in the
8:07project timeline and in the project
8:09features.
8:10Security, like ownership become mostly
8:12cosmetic. It exists on paper, but
8:14doesn't exist in reality.
8:16This is why you need security ownership
8:19and you need this authority, this
8:21ownership to come with authority.
8:23How to do that? A good model I've seen
8:26work for early stage company is for one
8:28of the founder to own security.
8:31They will decide how resources are
8:32allocated, how the product timeline is
8:35going to be aggressive. If a refactoring
8:37is needed, they will be able to say we
8:38are not going to go live until A, B and
8:41C is fixed.
8:43Now as you scale, you might be able to
8:45delegate that ownership and that
8:46authority to someone else and to create
8:48a structure for that. But many of the
8:51team that have good security posture
8:53today started with that model, sometimes
8:56even without without realizing it.
8:59So now you're in a situation where you
9:00know your risk and you re-evaluate them.
9:02Your exposure scale at a at a pace where
9:05you can control it and you have security
9:07ownership with clear authority. There is
9:09another aspect where I've seen sometime
9:12leader failing without even realizing it
9:14is how they communicate about security.
9:18Who here have been in a situation where
9:21you see a security report, there is a
9:23high security finding
9:25and the leader of the company or the the
9:28lead of like the protocol team
9:30is pushing back to reduce the severity.
9:33So you report high severity and now you
9:35enter kind of like a back and forth with
9:36the developer team to reduce the
9:37severity to medium or something like
9:39that.
9:40This creates
9:42a focus on the perception of security
9:45and not the security. It's it tells the
9:47engineer that the optics of how the
9:49report will looks like is more is more
9:51important that how the security issue
9:54might have impacted the team.
9:56I have been in a situation where I was
9:57reporting high severity to teams and I
10:00could see that the engineer was stressed
10:02because he was afraid of the consequence
10:04of having a high severity in this
10:06report.
10:07And this create again a situation and an
10:10environment where the perception of
10:12security is more important than the
10:13security itself.
10:15No, mature team behave really
10:17differently. The leaders want the
10:19assumption to be challenged. Where you
10:21do security review and an audit with a
10:23mature team, the engineer will seek
10:25feedback. They will not be defensive
10:27about their code. They will want to you
10:29to challenge the different assumptions
10:31that have been made within the code.
10:33When an engineer
10:36when an engineer see a concern, they are
10:38going to raise it internally. I am not
10:40confident about this component. I think
10:42we should spend more time.
10:44If a refactoring is judged necessary, it
10:47is going to happen. If it needs to delay
10:49a feature or a product launch, it is
10:51going to happen. This means that
10:53security is not just something that is
10:55for optics. It does have impact in the
10:58decision.
11:00I have been this is in a weird situation
11:03where I was reporting what I thought was
11:05a medium severity to a protocol team
11:08and the engineer started to argue with
11:09me about the severity, not to decrease
11:11it but to increase it. So I'm like this
11:14is medium and the engineer is like no
11:15no, this is high.
11:17This can only happen in an environment
11:19where the engineer feels safe to raise
11:22concern and uncertainty instead of
11:24trying to minimize it.
11:26This is why as a leader, the way you
11:28communicate with the team is really
11:29important. Whatever you say is going to
11:32have a large impact within your
11:33engineering team and you want your tone
11:36I want security to favor
11:40raising uncertainty, raising risk
11:42instead of trying to minimize it. So
11:44next time you receive a security report,
11:46if you see a high severity, do not focus
11:49on how we can make it look better for
11:51the end user. Focus on
11:53what can we learn from this from this
11:55finding. For example, ask your team
11:58what do you need? How can I help you? Do
12:01you need more time testing? Do you need
12:02more time thinking of design? What are
12:04the things we can change within our
12:05process for this to not happen instead
12:08of forcing focusing on security? Again,
12:10the optics might have some impact in
12:13your product, but in the long term, the
12:16if you focus more on the perception of
12:17security rather than security, you will
12:19not build an environment where concerns
12:21are properly addressed.
12:25So, it's 2026, we have to talk about AI.
12:29Some people believe AI will solve cyber
12:31security and AI will find all the bugs.
12:34This is not what we are seeing today. AI
12:37does decrease the cost of finding bugs,
12:39but everything we just discuss, risk
12:41management, perception of security,
12:44exposure scaling, um and security
12:47ownership, all of these still matter in
12:49the world where AI help us to find bugs.
12:53For critical code, I see many leader
12:57make the mistake of focusing on the AI,
13:00on how it can scale and make uh engineer
13:03write faster code. The question become
13:06how much can we ship, how fast can we
13:08ship.
13:09Uh and this is a problem because if your
13:12output and the capacity and if the
13:14number of line of code they are
13:15producing out weight your capability to
13:18manage risk and understand how your
13:20protocol growth and like the different
13:22limitation, you are more likely to have
13:24incidents.
13:27Now, as a leader, you can take a
13:29different strategy. You can take a
13:30different optic for critical code.
13:32Instead of focusing on how AI will make
13:35you faster, focus on how can you use
13:37this extra capacity to improve your
13:39security.
13:40This change how you measure the impact
13:42of AI within the company. You can know
13:45focus on okay, are we seeing a decrease
13:48of security findings during review? Are
13:51we using new technique like formal
13:52verification or fuzzing? Do we have more
13:55control over how we manage risk?
13:58Thinking of security
14:00as a first priority in how you integrate
14:01AI might uh might leads to increase
14:04speed and that might become a byproduct
14:06of this focus.
14:08But for critical code, you should focus
14:10first on how to improve your security
14:12before how you can be faster with AI.
14:15And this is again a leadership decision
14:17that will change how in general going to
14:19think and to to shape the code base that
14:21you are producing.
14:23So risk management, exposure scaling,
14:27security ownership, perception of
14:29security versus security itself. All of
14:32that, you know, comes down to how the
14:35leader think about security and do they
14:38consider security as something that
14:39shape decision or do they consider
14:41security as something that is a
14:43verification check at the end of the
14:45protocol development.
14:47And this is difficult sometimes for
14:48founder because they are not trained to
14:50reason this way.
14:51A good um
14:54a good parallel you can think of is
14:55profitability.
14:57As a founder,
14:58you will not think profitability is
15:00something you check at the end. Uh it's
15:02not even if you delegate the execution,
15:04it is still your responsibility.
15:06Profitability is going to shape how you
15:08are building your software, how you how
15:10aggressive you are, which feature you
15:13have, which limitation you are okay to
15:14accept or not. You can think of security
15:17in the same way.
15:19Now to build that mindset,
15:21there are a couple of simple questions
15:22you can ask yourself. The first one, if
15:25our TVL grows from 1 million to 10
15:28million, what additional security
15:30measure do we need?
15:31This means treating TVL growth not just
15:34as a success but as a risk vector.
15:38With a security budget, what level of
15:41complexity and what level of security
15:43can we responsibly support? This means
15:46not trying to retrofit your security
15:48budget to what you have built but you
15:50build something that match your security
15:52budget.
15:54The last two questions are related to
15:56risk management. So after an audit, what
15:59what stay possible and uh what
16:02assumption do we rely on and when do
16:04they stop to be true? So this is about
16:06knowing your risk and being having a
16:07process to reevaluate them over time.
16:13A lot of what we discussed today were
16:15around leader um but I would argue that
16:18anyone within their career can benefit
16:20from thinking of security as um
16:24something which has an impact within
16:25leadership. As an engineer, next time
16:28you look for a position, choose an
16:30environment where security matter, where
16:32the security leader, uh where the leader
16:34of the of the of the company you want to
16:36join consider security as something
16:38important. Ask question, you know, look
16:40for clues. If you care about security
16:43and the leader of the company you work
16:44with don't care about security, you will
16:47be frustrated. So take this as a
16:49consideration.
16:50For investor, here you have an important
16:53play.
16:54You can influence your portfolio company
16:56to think beyond we had an audit. Ask
16:58them question about the security
16:59ownership. Ask them question about how
17:01they manage risk. Um, you have a lot of
17:04power into how how to change their
17:05mindset.
17:07Finally, as a leader, make sure the
17:09security concern are going to allow you
17:11to change scope, timeline, and exposure.
17:15Remember the quote.
17:16We knew it was a weak point, but we
17:18thought it was okay. You do not want to
17:21be the next one to say that. Audit will
17:23find bug, but your leadership will
17:25determine if the condition that created
17:28the issue will improve or not.
17:30Thank you for your attention.
17:32[applause]
17:39Any question?
17:41>> Hey, I'm Fazeen building
17:43My question is what acceptance bar
17:46should leadership accept for a security
17:48fix before it ships?
17:49For example, today is just that proof of
17:52concept fails now. Is that enough?
17:55>> Um, so can you repeat the question?
17:59>> What acceptance bar should leadership
18:01accepts for a security fix before it
18:03ships?
18:04Today today is just that proof of
18:06concept fails now.
18:08>> Okay, is the question I want um if you
18:11receive a finding you want to have a
18:13proof of concepts uh
18:16uh okay I'm not sure to understand the
18:20question. Do you mean acceptance having
18:22you want to accept the how do you know
18:24you can accept the fees can ship
18:25basically? Okay, okay. So the question
18:27is when you receive a fix
18:29how confident can you be with respect to
18:31the fix, right?
18:33Okay.
18:34That depend on the complexity of the
18:36finding. If you have a finding and the
18:38fix is pretty straightforward, usually
18:39you can have a pretty good confidence
18:41that the fix will fix it. Now if the
18:43issue stem from something a bit more
18:45complex, you might need to ask your
18:47engineer
18:49do you think our issue might stem from
18:51that? For example, if the fix is just
18:53like
18:55like an access control missing, that's
18:57pretty easy to evaluate. If the fix
18:59require a change in the winding
19:00decision, that might require to you to
19:03allocate a bit more resource. Now as a
19:04founder, if you're a bit unsure about
19:06that, I will ask the engineering team
19:08how they think about it because your
19:09engineer knows the protocol better than
19:11anyone else. You can also ask your
19:13security you know expert that internally
19:15or externally how they think about it.
19:17But asking the engineering team is
19:19probably the first place to to to start
19:21with.
19:25>> Hello.
19:26Great talk. Thank you, Jocelyn. So
19:28question I have is about incidents and
19:31incidents response. I feel like it's one
19:33of the scariest things for leaders and
19:36founders and whatever
19:38especially in our crypto companies to
19:40have an incident, but it feels like it's
19:42inevitable nowadays. And so how do you
19:45how do you make it part of I guess
19:48regular day-to-day in business to handle
19:50an incident, have an incident response
19:52and all that sort of stuff?
19:54>> Okay, that's a really good question.
19:56So yeah, you have to prepare for
19:58incident because it is likely to happen.
19:59For that you can think of it ahead of
20:01time. For example, how you monitor your
20:03code base not as the code base is
20:06written, how we are going to monitor it,
20:07but we are going to build like this new
20:09component from the design phase. How can
20:12we integrate monitoring there? What do
20:13we need to monitor?
20:15You can test your incident response
20:16there like war game and and think like
20:18that. The SEAL which is a Security
20:20Alliance has a framework website which
20:23has a lot of different kind of like
20:25advice on how to do it. So I think not
20:28considering incident as
20:30a last step within your development but
20:33something you integrate from the get-go
20:35will help you and testing it is also
20:37really important because you don't want
20:38to be in a situation where
20:40you think you have good monitoring but
20:42actually you don't. Testing monitoring
20:44is also really difficult because
20:47depending on the setup you have testing
20:49like the tooling you are using might be
20:50tricky so you want to get that ahead of
20:52time. There are a couple of companies
20:54that focus on that that can also help
20:55you you can you know reach out to them.
21:00>> Hello Jocelyn great great talk. So this
21:03question is more from the leadership
21:04side like towards the leadership. So as
21:07we know like the unfortunate reality is
21:09many of the protocols they still follow
21:11the regulations which are already out
21:12there. However the the velocity of the
21:15incidence that we are seeing that is
21:17overriding the regulations that are
21:18already put into place. So while you are
21:21mapping out let's say some incident that
21:22okay a new attack pattern has appeared
21:25you don't have the regulations for it
21:26and people are like that okay because we
21:28don't have the regulations for it maybe
21:29we not we we are not required to plan
21:31for it but you will still have like
21:33incurred losses.
21:35So how as we as a community ensure that
21:37the regulations are also you know they
21:40are like they have a velocity as well
21:42which is matching up the incidents. I
21:43think that is really important.
21:45>> Okay
21:46that's a good question and a difficult
21:47one.
21:49I don't think we should wait for
21:52regulation to tell us what to do.
21:54Regulation might help us to have some
21:56baseline but it is not going to be
21:59enough and you know we have seen that in
22:02other industry when when we start having
22:04like some certification and think like
22:06that
22:07it can backfire because usually baseline
22:11from regulation are not that high and
22:14they create kind of a floor for all the
22:16company to follow but that doesn't push
22:18company to go beyond that and that's why
22:21building a security mindset for a leader
22:24is important because if the leader think
22:25of okay, there is this framework from
22:28like the regulator, we follow it, we are
22:30good, that is not going to be enough and
22:32that's not going to be sufficient. So
22:34they might help a bit, but I would not
22:36focus on them.
22:40>> Um whenever you're dealing with consumer
22:42protection, um
22:45uh for example, like you might care a
22:47lot about your security and making sure
22:48that your systems are not hacked, but
22:50consumers can lose credentials, they can
22:52be tricked into signing into
22:54applications you have no control over.
22:55How should should security leadership
22:58think about protecting consumers even if
23:00it goes kind of beyond the scope of the
23:02organization and how do they how do
23:05should like security leaders try to get
23:06other peoples in the organization to
23:08care about it even if from their
23:10perspective it's going out of their way
23:12to spend resources on things that are
23:14maybe not their immediate problem?
23:16>> Okay. Um
23:18so like the question is about how how
23:21leader can think of
23:22security for the end user which lie into
23:25like the limit of the protocol itself,
23:28right? So if it's if it's like on your
23:30off-chain component like your website
23:32and think that are connected to your
23:33website, I would argue it's still within
23:36your responsibility, right? Now the
23:38issue is where you have like integration
23:40with other protocol and when you have
23:42like
23:42the limit is a bit more gray.
23:45Um this is where coordination among the
23:48industry is important and discussion
23:50among like the different important
23:51player is important because you as a
23:54leader you will be able to secure
23:57or try to secure your protocol. You
23:59won't be able to create and enforce like
24:02some additional mitigation and safeguard
24:04outside of your organization, but you
24:06can try to discuss with a different
24:08partner and with a different provider to
24:10push in them in that direction.
24:12Um if you share common investor with
24:15like other other other company you work
24:18with, you might talk to your investor
24:19and try to have like you know a natural
24:21ground to push some you know mitigation,
24:24some something. So try for for example
24:26to find some natural party that can help
24:29you to push
24:30you know some additional security
24:31measure.
24:45Yeah.
24:46>> say you're
24:47a smaller stage company, you know you're
24:49not an Aave, Morpho, Uniswap
24:52and you want to communicate. You take
24:54security very seriously as you, you know
24:57discussed in this presentation of how
24:59you need to have an ownership mindset
25:01and to be able to you know
25:04take responsibility. We're not just
25:05going to push it on the auditors.
25:07How do you then convey that to the
25:09public if you don't have
25:12you know the resources to get 10 audits,
25:14you know to kind of prove that?
25:16>> Okay, that's a really good question.
25:18How how can you convey that you have a
25:19security mindset
25:21if you cannot pay for you know
25:2410 audits, right?
25:26You you can share some of the things you
25:28are doing like concrete example of what
25:30you are you are doing to do that. For
25:32example,
25:33maybe you use fuzzing, maybe you use
25:35like you know formal formal verification
25:37within your codebase. Maybe you know you
25:39dedicate some resource time from like
25:41your engineer to look for vulnerability
25:44in other codebase or learn from past
25:46exploit. So you have way to communicate
25:48concrete example of what you are doing
25:51to help like the security of your
25:52protocol and you can can communicate
25:54that.
25:55So I will look I will look for concrete
25:57example that can be translated into
25:59something public.
26:02Yeah.
26:06>> [clears throat]
26:15>> Thanks.
26:16With Aave as an example, how do you
26:19consider
26:21I guess
26:22it's
26:23the wider scope of say hacks on
26:27cred protocol that end up in Aave
26:30and and you something that's so far out
26:33of your scope in many ways.
26:35>> Yeah, that's a good question,
26:38because you know, some protocol like
26:39Aave and other,
26:41we are impacted when an incident happens
26:44outside of organization, right?
26:47And this is scary because we can do as
26:49much as we can within organization,
26:51similar to the previous question, but we
26:53cannot control what's happening outside.
26:55So it's really about monitoring, it's
26:57about evaluating who you're interacting
26:59with. So in the case of Aave, we have
27:01like, you know, a risk framework, we
27:02have like a lot of different
27:03consideration and and additional
27:06validation step that we are doing for
27:08the people we interact with and the
27:09assets we're onboarding.
27:11So it's really about like pushing what
27:14you know and trying to one, evaluate who
27:17you're interacting with, and two, if you
27:19have the resource, trying to help them
27:21to scale up their security.
27:29Okay, thank you.
Source and accuracy details
- Names, organizations, technical terms, code and numbers may be misrecognized.
- Source wording and disfluencies are preserved; no editorial corrections were applied.
- Export contains whole-second cue starts only; cue ends and subsecond timing are unavailable.
- A full caption-track export does not establish that every spoken word is represented.
- Automatic YouTube captions; not human-verified against the audio.
Source export SHA256: 2e8205e1daa14ce3ddfff68c76f91ba5ed000f06dd55fd5f576c0e1bcdf85d26