[00:08.280 --> 00:14.800] Hello, it's nice to be here. Thank you for joining this session. I'd like to talk today [00:14.800 --> 00:21.040] about a policy approach to resolving cyber security problems in the election process. [00:21.240 --> 00:27.620] One thing's for sure, after the coronavirus is no longer dominating the news, [00:27.620 --> 00:35.380] election security will come back to center stage, big time. It's not a complicated subject that [00:35.380 --> 00:41.000] it is a complicated subject that few people really understand, even election officials. [00:41.940 --> 00:46.580] So today, I want to talk about the role that private sector companies play in the voting [00:46.580 --> 00:53.100] system, the vulnerabilities associated with their role, and discuss a path forward from the legal [00:53.100 --> 01:08.320] and policy perspective. So election security, it's really a private sector problem. [01:09.170 --> 01:19.340] Election security, so many aspects of it are performed by private sector companies. [01:19.340 --> 01:25.600] There are many avenues for tampering with an election, including changing votes, [01:25.600 --> 01:29.840] causing machines to malfunction, altering voter registration records, [01:29.840 --> 01:35.260] and disrupting equipment to be used that's used to check in voters. [01:35.960 --> 01:42.400] The private sector companies play a large role in these activities, which we call the election [01:42.400 --> 01:50.300] process. The voter registration, the checking in, the voting, the polling, the tabulating of the [01:50.300 --> 01:58.800] votes, all that is called the election process. And the private sector companies sell the voting [01:58.800 --> 02:05.960] machines and program them for most elections. Yes, they program them for each individual election [02:05.960 --> 02:13.400] in most jurisdictions. They register voters, they tally votes, and they report votes. [02:13.400 --> 02:19.680] And there are vulnerabilities associated with their role that most people don't understand. [02:20.220 --> 02:25.880] They think that this is all something controlled by state and local election officials, and it is [02:25.880 --> 02:36.600] not. So today I want to talk about a path forward and how to solve this huge complicated problem. [02:38.090 --> 02:39.200] There are [02:41.310 --> 02:49.240] many avenues for tampering with an election, including changing votes, causing machines [02:49.240 --> 02:55.620] to malfunction, altering voter registration records, and disrupting equipment used to check in voters. [02:57.140 --> 03:04.380] Let's look at the vulnerabilities in the voting equipment. In 2005, Hari Hirstie performed the [03:04.380 --> 03:10.520] famous Hirstie hack and successfully altered votes in a one-step hack that changed both [03:10.520 --> 03:18.380] central tabulator results and the voting machine results tape. It was the digital equipment of [03:18.380 --> 03:25.000] stuffing the ballot box. The election official who invited Hirstie to check the de-bolt acu-vote [03:25.000 --> 03:31.520] optical scan voting machines said he wouldn't have been able to detect the change and would [03:31.520 --> 03:39.340] have certified the election. De-bolt is now owned by Dominion Voting Systems. None of the [03:39.340 --> 03:46.820] vulnerabilities found by Hirstie were ever fixed. And these same machines are planned for use in 20 [03:46.820 --> 03:55.980] states in the 2020 election. Think about that. It's 2020 and he discovered this in 2005, [03:55.980 --> 04:01.980] 15 years later. That's crazy. A later model of the same voting machines with the same [04:01.980 --> 04:09.060] vulnerabilities was used in a hotly contested and disputed 2016 election between Stacey Abrams and [04:09.060 --> 04:18.320] Brian Kemp in Georgia. So many of these systems that are using these machines are using machines [04:18.320 --> 04:26.000] with these vulnerabilities. In recent elections, 99 percent of votes in the U.S. were cast or [04:26.000 --> 04:34.000] counted on computers. And many of the core election systems, voter registration databases, [04:34.000 --> 04:42.820] election management systems, voting machines, and vote counting systems are using aged computer [04:42.820 --> 04:49.260] equipment. The systems employ software that can no longer be updated or patched. [04:49.260 --> 04:54.520] They include databases that have known vulnerabilities and they're managed by [04:54.520 --> 05:01.280] third-party vendors where supply chain risks exist. This is a big problem. [05:06.560 --> 05:11.540] Let's look at the cyber attacks on these private vendors. [05:12.360 --> 05:22.260] In 2016, Gresha's military intelligence service penetrated VR systems. VR systems is the vendor [05:22.860 --> 05:28.360] that manages and handles all of the programming for the majority of the counties of Florida [05:29.360 --> 05:33.340] and it handles all absentee ballots and early voting. [05:34.300 --> 05:44.400] We know of this penetration because an NSA contractor, Reality Winner, [05:44.400 --> 05:52.740] released a top-secret report about it. And we later found out the FBI briefed election officials [05:52.740 --> 06:02.440] in Florida on a very secret basis. It was a year later that election officials around the country [06:03.080 --> 06:08.280] realized that Gresha's military intelligence service had been penetrating [06:09.240 --> 06:16.720] VR systems and that perhaps their systems might also be vulnerable. In the 2016 elections, [06:16.720 --> 06:22.960] electronic voter ID systems in several states went down in certain precincts. [06:22.960 --> 06:27.160] Now these are the systems that help identify a voter when they come in to vote. [06:27.680 --> 06:34.680] And these technical glitches, they called them, in the machines called hours-long waits [06:34.680 --> 06:41.960] for people who had come to the polls to vote. Hours-long waits. Some voters [06:43.900 --> 06:51.140] were unable to wait and others could not vote before the polls closed. [06:52.020 --> 06:58.840] On election day in 2019, federal officials from law enforcement, homeland security, [06:58.840 --> 07:05.220] and the intelligence community issued a joint statement declaring that, [07:05.220 --> 07:12.040] our adversaries want to undermine our democratic institutions, influence public sentiment, [07:12.040 --> 07:19.240] and affect government policies. Russia, China, Iran, and other foreign malicious actors [07:19.740 --> 07:26.680] all will seek to interfere in the voting process or influence voter perceptions. [07:29.080 --> 07:39.160] Wow. And it's true. We have a serious problem. We have governments from outside the U.S. [07:40.000 --> 07:47.900] trying to influence our democratic processes. And we have voting machines and voter ID systems [07:48.120 --> 07:57.600] and private sector companies that have not presented any level of security and assurance [07:58.080 --> 08:03.940] that their processes and systems have integrity and are confidential. [08:08.250 --> 08:21.330] So we have a U.S. Election Assistance Commission. And this commission is supposed to, [08:21.330 --> 08:26.450] but it's perhaps the weakest link in the nation's voting system. [08:26.450 --> 08:34.690] The EAC is a bipartisan commission established by Congress in 2002. It maintains a national [08:34.690 --> 08:42.310] mail voter registration form. It accredits testing laboratories and certifies voting systems. [08:42.310 --> 08:47.810] And it serves as a national clearinghouse of information on election administration. [08:48.810 --> 08:56.040] In December 2016, Recorded Future reported that a Russian-speaking hacker named Rasputin [08:56.470 --> 09:10.370] was selling access to the EAC systems on the internet. Rasputin had full admin access to [09:10.370 --> 09:16.890] the database and could upload any file he wanted. He had lists of voting machinery, [09:16.890 --> 09:23.890] test reports of their software, and knew where they were deployed. An EAC employee, [09:23.890 --> 09:30.150] whose credentials had been compromised, said if Rasputin had access to the database, [09:30.150 --> 09:34.310] he could access the server where the proprietary information is kept. [09:35.150 --> 09:42.410] The EAC keeps information about vulnerabilities in voting systems. Thus, a hacker who gets into [09:42.410 --> 09:50.010] the EAC could find out where the weak links are in the voting systems all around the country. [09:59.200 --> 10:05.500] So let's look at the voting machine companies and what role they have played in this. [10:06.320 --> 10:15.540] They have been recalcitrant and arrogant. Researchers and cyber experts have found [10:15.540 --> 10:21.700] multiple vulnerabilities in the most used voting machines that would enable an attacker [10:21.700 --> 10:26.800] to gain full access to a system, change configurations, and install a modified [10:26.800 --> 10:35.120] operating system without election officials knowing. These vulnerabilities can enable [10:35.120 --> 10:42.300] hackers to change an election, shut the system down, enable remote execution of code, [10:42.300 --> 10:49.520] and offline ballot tampering. There are three primary vendors for voting machines. [10:49.560 --> 10:57.560] Electronic systems and software, known as ES&S, Dominion Voting Systems, and Hart InterCivic. [10:58.140 --> 11:02.560] We know there are vulnerabilities in most of the voting machines, [11:03.060 --> 11:10.600] but very little is known about the security of these companies' own IT systems. [11:11.200 --> 11:18.700] The companies that produce these voting machines, that are supposed to maintain them and produce [11:18.700 --> 11:26.340] them, and they are the underpinning of the core in our election process. Very little is known [11:26.340 --> 11:33.320] about the security of the companies that even manufacture these machines. Some voting machines [11:33.320 --> 11:41.020] are optical scanners, some have touchscreen voting, some use QR codes or barcodes, [11:41.020 --> 11:47.700] and others send votes in clear text back to vendors to be tallied. They can all be hacked [11:47.700 --> 11:56.460] or compromised. If almost all the voting precincts in our clunky system use equipment from these [11:56.460 --> 12:02.100] vendors, an attacker only needs to hack the equipment to reach all of the voters. [12:03.320 --> 12:11.040] Unlike major technology companies, such as Apple and Microsoft, these vendors do not allow [12:11.040 --> 12:18.660] researchers to test their equipment and review their code to find vulnerabilities and bugs. [12:18.720 --> 12:25.700] The symbiotic relationship between tech software and hardware vendors and researchers helps them [12:25.700 --> 12:30.200] improve their products and keep them secure. But this is not happening [12:30.860 --> 12:39.460] when it comes to voting machine manufacturers. They will not let the researchers have access [12:39.460 --> 12:46.660] to their equipment. Voting equipment companies have been highly resistant to any review by the [12:46.660 --> 12:53.680] research community, claiming their systems are safe and secure. Yet they have failed to fix [12:53.680 --> 13:01.500] identified vulnerabilities in the voting equipment. They repeatedly claim everything [13:01.500 --> 13:07.500] is fine, we're all secure, this is a priority, we take this seriously, America's voting is just [13:07.500 --> 13:15.660] our basic concern, and they don't fix vulnerabilities that researchers have found [13:15.660 --> 13:25.900] for 15 years. They let their voting machines be out front and used for voting knowing they [13:25.900 --> 13:31.640] have vulnerabilities in them. I guess they think no one's going to exploit them. [13:32.060 --> 13:38.480] But really, to cover that up and try to say they're all safe and sound, that's just wrong. [13:41.310 --> 13:46.150] So during the three years of the voting village's existence, I think most of you are familiar with [13:46.150 --> 13:52.150] voting village, none of the vendors have supported the effort, nor have they been willing [13:52.150 --> 14:02.850] to donate or offer equipment. While the researchers in the voting village, they're dedicated to this [14:02.850 --> 14:09.830] because one, they want to help develop a community of cybersecurity experts in election security. [14:09.830 --> 14:14.770] There are not very many experts out there who know how to [14:15.610 --> 14:21.430] deal with election voting machines and cybersecurity problems and how to solve [14:21.430 --> 14:28.290] those vulnerabilities. And they also want to make the voting machines and equipment more [14:28.290 --> 14:37.210] secure by letting these vulnerabilities be known. The problem is the vendors are doing nothing about [14:37.210 --> 14:46.430] it. And apparently the election officials also are not requiring them to fix these vulnerabilities [14:46.430 --> 14:55.690] before they have another election. So that's a big problem. So I want to put forth a proposal [14:56.470 --> 15:03.030] to address this problem and achieve results. Certain actions can be taken at the federal level [15:03.700 --> 15:08.310] that will push a standardized approach out to state and local election officials [15:09.220 --> 15:17.670] and will require certain actions to be taken by private sector companies. [15:19.170 --> 15:26.490] Article 1, section 4 of the U.S. Constitution grants Congress the power to regulate the times, [15:26.490 --> 15:34.470] places, and manner of holding federal elections. Federal elections deal with elections of senators, [15:34.470 --> 15:41.030] representatives, and the president. Now state and local election officials are responsible for [15:41.030 --> 15:47.590] conducting all elections, but they depend on infusions of federal funding to supplement [15:47.590 --> 15:54.530] their state and local funding. So state and local officials can't really afford to have [15:54.530 --> 15:59.570] separate equipment and systems, one for state elections and one for federal elections. [15:59.630 --> 16:06.830] So when federal election officials or federal election Congress mandates certain requirements [16:06.830 --> 16:12.490] for federal elections, they pretty much have to go along with those because they only have one [16:12.490 --> 16:20.770] system for voting. So therefore, if Congress sets requirements for federal elections [16:21.270 --> 16:26.930] and restricts the funding that these state and local election officials need [16:27.630 --> 16:35.630] to only those election agencies that adhere to federal requirements, we will begin to see [16:35.630 --> 16:44.430] consistent actions taken across the U.S. that will tighten cybersecurity in the election process. [16:45.610 --> 16:56.330] So this proposal that I want to go over with you today is such a proposal to have [16:57.650 --> 17:05.310] federal requirements set by Congress. The first would be to direct NIST, the National Institute [17:05.310 --> 17:14.210] of Standards and Technology, to establish federal standards for cybersecurity. NIST is [17:14.510 --> 17:20.170] the entity in the government that has established all the federal information processing standards. [17:20.170 --> 17:26.370] It's established all of the cybersecurity best practices and standards that have been put forth [17:26.370 --> 17:34.750] by NIST that federal agencies have to adhere to and certain federal contractors. So they have a [17:34.750 --> 17:42.810] deep bench of expertise in not only standards and cybersecurity but in secure engineering practices. [17:43.200 --> 17:50.090] And it would be appropriate to direct them to develop federal standards for cybersecurity of [17:50.090 --> 17:59.710] our election system software, the infrastructure, and the hardware. All three levels that are used [17:59.710 --> 18:09.010] in voter registration, vote tallying, voter polling, and in the manufacturing, servicing, [18:09.010 --> 18:13.250] or writing of election parameters of voting machines and equipment. [18:15.530 --> 18:22.070] The first covers the systems, the infrastructure, the hardware, the software, the hardware, [18:22.070 --> 18:28.010] the network that they're using. What are the standards for that? Because that is used in [18:28.010 --> 18:34.590] registration, in tallying, in voter polling. Again, this isn't just one thing of one machine [18:34.590 --> 18:40.670] and go to the vote, the poll and vote. This is registering to vote. It's signing in and being [18:40.670 --> 18:49.130] ID'd. It is voting. It's tallying the votes. It's polling the votes, the voters, and it's reporting [18:49.130 --> 18:57.810] the votes. All of these actions are largely performed by private sector companies who we [18:57.810 --> 19:05.610] don't have any idea how their cybersecurity program is, whether it's mature or not. [19:06.230 --> 19:12.930] We suspect it's not very mature, but that's just a suspicion. But we should have a standard for [19:12.930 --> 19:18.630] saying it must meet these standards. Our democracy depends on it. That's worth a standard. [19:19.730 --> 19:25.250] And then we have in the manufacturing, the servicing of voting machines and equipment. [19:25.250 --> 19:32.010] Absolutely. We want these vulnerabilities fixed. They need to be serviced and maintained and have [19:32.010 --> 19:37.350] integrity. The writing of election parameters means the programming of these machines. [19:37.350 --> 19:44.450] In most of the jurisdictions, the private sector companies program these machines for [19:44.450 --> 19:51.510] every single election. We need standards to govern how that's done. Second is we should [19:51.510 --> 19:59.770] direct NIST to establish a certification process for the security and integrity of election systems, [19:59.770 --> 20:05.930] software, infrastructure, and hardware, and associated components and modules of the election [20:05.930 --> 20:13.230] process. So there should be a certification process to make sure they're meeting the standards. [20:13.610 --> 20:19.190] Next, we should direct NIST to analyze the private sector's role in the election process [20:19.630 --> 20:25.370] and recommend any roles or functions that should be changed or restricted to public sector [20:25.370 --> 20:33.190] election officials. There are some roles being undertaken by private sector companies that [20:33.190 --> 20:42.990] perhaps should not be a private sector activity. Perhaps that role should be strictly a governmental [20:42.990 --> 20:50.650] function. But let's NIST analyze that in all of their work in establishing the federal standards [20:50.650 --> 20:58.850] in the certification process and recommend to Congress roles that should be perhaps reserved [20:58.850 --> 21:08.950] to the public sector. Next, we want Congress to pass a law restricting federal funding [21:09.910 --> 21:15.170] to election jurisdictions to only those jurisdictions that one [21:15.830 --> 21:22.570] use such funding in a manner consistent with the NIST federal election standards. [21:23.150 --> 21:29.890] And those jurisdictions that require annual cybersecurity assessments by an independent [21:29.890 --> 21:35.910] third party of all the systems in the election process in accordance with the standards [21:37.450 --> 21:44.390] that require annual cybersecurity assessments by a third party of all private sector companies [21:44.390 --> 21:49.870] involved in the election process in accordance with the NIST standards. And to make these [21:49.870 --> 21:55.670] assessments available to election officials contracting with them. A company may get a [21:55.670 --> 22:02.590] cybersecurity assessment but it may not share it. We want them, we want these vendors to have to get [22:02.590 --> 22:11.050] third-party assessments every year and share those assessments with the election officials [22:11.050 --> 22:16.230] contracting with them. And we want the election officials to get third-party assessments of [22:16.230 --> 22:24.170] their systems in accordance with the NIST standards. This is what private businesses do [22:25.010 --> 22:29.450] every day. That's what's required of them. This is not asking too much [22:30.110 --> 22:33.950] of election officials or private sector vendors that support the democracy that [22:33.950 --> 22:40.910] this country has been built on. We also want federal funding restricted to those jurisdictions [22:40.910 --> 22:47.510] that establish requirements for post-election auditing of votes, at least on the level of [22:47.510 --> 22:55.190] risk-limiting audits and to make the findings public. So restricting federal funding to only [22:55.190 --> 23:00.750] those jurisdictions that comply with the NIST standards, that conduct the risk assessments [23:00.750 --> 23:06.930] themselves, that make their vendors get the assessments, and that have post-election auditing. [23:06.970 --> 23:15.390] There's a precedent for this. Congress has passed laws in the past to restrict highway funds [23:15.390 --> 23:20.230] to only those jurisdictions that lower the speed limit to 55 miles an hour. [23:21.190 --> 23:28.170] Congress has restricted funds to only go to those school districts that would adjust their [23:28.170 --> 23:36.360] cafeteria menus to comply with the new recommended federal menu. There are numerous other examples, [23:37.050 --> 23:45.470] but the Congress has in the past, in several instances, tied its federal funding to meeting [23:45.470 --> 23:52.130] certain requirements. There's nothing more important than having our federal funding tied [23:52.130 --> 23:58.610] to meeting the requirements that our vote counts, that every vote should be counted and it should [23:58.610 --> 24:07.370] be counted as cast. The American Bar Association, which represents over 400,000 attorneys, recently [24:07.370 --> 24:17.850] adopted Resolution 118, which calls for these exact measures. Cybersecurity best practices and [24:17.850 --> 24:23.850] standards can help, because election security is not going to get solved overnight. But standing [24:23.850 --> 24:31.370] up a cybersecurity program for election systems that's in compliance with cybersecurity best [24:31.370 --> 24:38.710] practices and standards will be a first big step. Congress may not pass a law right away, [24:38.710 --> 24:44.750] but election officials should already be doing this. They should already be saying, [24:44.750 --> 24:50.150] we have NIST standards for cybersecurity programs. There are ISO standards. There's [24:50.150 --> 24:54.990] multiple standards out there for cybersecurity programs, and every single vendor should be [24:54.990 --> 25:00.190] adhering to those. They're not only using equipment with vulnerabilities, their own [25:00.190 --> 25:06.370] networks and systems are vulnerable as well. So it's time election officials step up and take [25:06.370 --> 25:13.070] action, even before Congress does. They should require, election officials should require election [25:13.070 --> 25:18.630] vendors to meet cybersecurity standards and best practices, conduct annual risk assessments of [25:18.630 --> 25:25.610] their program's maturity, and do what every other business would do, manage its risk. [25:25.610 --> 25:30.810] In addition, they need to follow the lead of California and other jurisdictions and hire [25:30.810 --> 25:36.650] experts to test the security of their voting machines and equipment, and then demand that [25:36.650 --> 25:43.990] vendors close any vulnerabilities found. Why are these officials letting machines with vulnerabilities [25:43.990 --> 25:50.550] that have been known for 15 years be used in their jurisdictions? That's crazy. They need to demand [25:50.550 --> 25:57.330] these vulnerabilities be fixed, and they need to join with each other nationally and ensure that [25:57.330 --> 26:04.410] that those machines are not used in elections. So legislation and other reforms are needed, but [26:04.410 --> 26:09.850] election officials can achieve these things through their own direction and in legal agreements with [26:09.850 --> 26:16.700] their election vendors, and they should begin now and do as much as possible before November. [26:17.560 --> 26:19.120] Thank you very much.