Government election websites in the United States drew website vulnerability attempts 180 times more often than denial-of-service attacks during the 2026 midterm cycle, according to an analysis Cloudflare published today. The company, which says it helped protect more than 1,400 election-related websites through three free programs, found that attackers spent the year probing state and local voting sites for a way into internal systems and data rather than trying to knock them offline. On November 3, Election Day, it recorded no significant distributed denial-of-service (DDoS) attack across any of those sites, while ordinary traffic to state and local election websites rose to about 602 million requests.

The post, written by Patrick Day, Jocelyn Woolbright and Carlos Azevedo, puts figures on the shift. Vulnerability exploit attempts climbed from 200,000 a day in January 2026 to 2 million a day in October, peaking in late October, several weeks before the vote. Over the 12 months before the election, Cloudflare says it mitigated 392 million such attacks on government election websites alone. And on October 22, one day after a critical PHP flaw became public, a state government election website in the Western United States set off Cloudflare's PHP defense more than 9,000 times.

In Short

Cloudflare says the US government election websites it protected in the 2026 midterms faced attempts to break into their systems 180 times more often than attempts to knock them offline, and that Election Day passed without a single significant DDoS attack. That matters to anyone who runs election, campaign or news websites, because the pressure built for months before the vote, reaching 2 million exploit attempts a day in October, while newsrooms and civil society groups still absorbed the bulk of the flood attacks. What changes is the picture of election-season risk: on the day itself, the biggest load came from voters, with requests to state and local election sites more than four times higher than two days earlier.

Voters, not attackers, filled the servers

"Ordinary user traffic to state and local government election websites was higher on Election Day (November 3) than any other day in 2026," according to Cloudflare. Requests to those sites reached approximately 602 million, up from 130 million on November 1, a Sunday. By PPC Land's calculation, that is 4.6 times the Sunday level, an increase of 363%, which matches the figure Cloudflare puts at the top of its post. Traffic across the rest of Cloudflare's global network rose by only 12% over the same period, according to the company.

There is a wrinkle in the wording. The post's opening sentence says "visitors" to government election websites "surged 363%," while the section that gives the underlying numbers counts requests. A request is not a person: a single visit to a polling-place lookup can generate many requests for pages, scripts and images, and the post gives no count of unique visitors.

Why these sites draw attention is set out plainly in the post. "State and local government election websites provide authoritative information to voters on how to register, where to vote, and who won," according to Cloudflare. "Because of their critical role, these platforms are often targeted by those seeking to suppress voter turnout, disrupt the reporting of results, or steal sensitive information."

Cloudflare's vantage point is large. The company's investor slides, covered when it reported third-quarter results on November 5, say 20% of the internet and 37% of the top 10,000 websites run on Cloudflare, and the company described its Radar data in October as drawn from roughly a fifth of all web traffic. Its 2025 Year in Review put global internet traffic growth at 19% for that year.

The figures in one place

Measure Figure Scope and period, according to Cloudflare
Requests to state and local government election websites about 602 million Election Day, November 3, 2026
Requests to the same sites 130 million November 1, 2026
Increase, Election Day vs November 1 363% Described in the post's lede as "visitors"
Traffic across the rest of Cloudflare's network up 12% Same period
Election-related websites protected more than 1,400 Athenian Project, Cloudflare for Campaigns, Project Galileo
Website vulnerability attempts vs DDoS attacks 180 times more often Government election websites, 2026 cycle
Vulnerability attempt rate vs Cloudflare's broader customer base roughly 10 times Government election websites, 2026 cycle
Vulnerability exploit attempts per day 200,000 January 2026
Vulnerability exploit attempts per day 2 million October 2026
Website vulnerability attacks mitigated 392 million Government election websites, 12 months before the election
Significant DDoS attacks 0 Election Day, across 1,400 election-related websites
The only significant DDoS attack of the period 520,000 requests November 2, 2026, a US House candidate in California
Malicious share of emails to Cloudflare for Campaigns participants 3.9% 90 days before the election
Malicious share across other Cloudflare customers 4.8% Baseline
Malicious emails that bypassed basic email authentication checks nearly 42% 90 days before the election
DDoS requests mitigated against journalism and election-related civil society over 10.8 billion 12 months before the election
Largest single attack 3.04 billion requests September 10, 2026, a US-based news organization
DDoS share of all attacks on Project Galileo entities 81% 120 countries, report released "earlier this year"
PHP defense triggers more than 9,000 October 22, 2026, one state government election website, Western US
Moldovan Election Commission, comparison more than 898 million malicious requests in 12 hours, peak 324,333 requests per second Election Day, 2025
Election Day requests vs November 1 (PPC Land calculation) 4.6 times 602 million / 130 million
Daily exploit attempts, October vs January (PPC Land calculation) 10 times 2 million / 200,000
Average daily vulnerability attacks mitigated (PPC Land calculation) about 1.07 million 392 million / 365 days

From floods to break-ins

The central finding is a change in what attackers wanted. "Unlike what we saw in other recent elections, attackers focused on gaining access to internal systems and data, rather than knocking websites offline," according to Cloudflare. Government election websites faced vulnerability attempts at roughly ten times the rate of Cloudflare's broader customer base, the company says.

What does such an attack look like? The post gives two examples. "Website vulnerability attacks exploit outdated or vulnerable website code to gain access to internal systems and data," according to Cloudflare. "For example, cross-site scripting attacks inject malicious code into web pages in order to capture administrator credentials or deface site content. SQL injection manipulates database queries allowing attackers to potentially extract voter records or alter data."

The volume rose through the year. From 200,000 attempts a day in January, the count reached 2 million a day in October, a tenfold increase by PPC Land's calculation, and peaked in late October, several weeks before polling day. The 12-month total of 392 million works out to an average of about 1.07 million a day (PPC Land calculation), which sits between the two monthly rates the post gives.

The post uses several labels for this category: "vulnerability scans," "vulnerability attempts," "vulnerability exploit attempts" and "website vulnerability attacks." It does not separate automated scanning, which looks for weaknesses, from attempts to use a weakness once found. Nor does it say whether any attempt succeeded.

Automated probing of this kind is not confined to election offices. In August, GreyNoise tracked scanners on 824 IP addresses forging the names of AI crawlers while hunting for credential files, and on November 16 an FBI alert tied more than 1,000 infected WordPress sites to a malware operation linked to over $28 million in ransomware payments. The first shows how routine automated probing of websites has become; the second, what a compromised site can be turned into.

One flood attack worth counting

Election Day itself was quiet on the DDoS front. "On Election Day itself, Cloudflare did not record a single significant distributed denial-of-service (DDoS) attack across any of its 1,400 election-related websites," according to the post. The only significant DDoS attack Cloudflare reports came the day before, on November 2, against a candidate for the US House of Representatives in California, and totaled 520,000 requests. The post names neither the candidate nor the district, and it does not say whether the site went down.

Cloudflare sets that number beside Moldova. The Moldovan Election Commission experienced more than 898 million malicious requests in a single 12-hour period on its Election Day in 2025, with attacks peaking at 324,333 requests per second, according to the company. By PPC Land's calculation, the Moldovan 12-hour total was roughly 1,700 times the size of the California attack.

Earlier Cloudflare figures mark the upper end, measured in bandwidth rather than requests. In October 2024 the company disclosed a DDoS attack that peaked at 3.8 Tbps, then the largest publicly disclosed, and its March 2026 threat report described a 31.4 Tbps baseline for DDoS traffic, which it said was fueled partly by the Aisuru botnet, with hyper-volumetric incidents up 700% in 2025. The election post gives request counts only and does not define what makes an attack "significant," so the threshold behind the Election Day zero is not stated.

Phishing that passed the first checks

Campaigns protected through Cloudflare for Campaigns received malicious email at about the same rate as everyone else. In the 90 days before the election, 3.9% of emails sent to participating organizations were identified as malicious, against a 4.8% baseline across other Cloudflare customers, according to the post.

The difference lay in quality. "Nearly 42% of those malicious emails managed to bypass basic email authentication checks, such as verifying sender identity or preventing in-transit tampering," according to Cloudflare, which says its advanced phishing detection tools still identified them. Applied to the 3.9% rate, that means roughly 1.6% of all mail reaching participating campaigns was malicious and had passed those basic checks (PPC Land calculation).

The company's broader data offers a related figure. Cloudflare's 2026 threat report, published in March, found that nearly 46% of analyzed emails failed DMARC, one of the main email authentication standards. The two numbers measure different things: one is the share of all analyzed email that failed a check, the other the share of malicious email that passed basic checks.

Newsrooms absorbed the floods

The picture flips for the press and civil society. "Although website vulnerabilities were the most common attack overall, for journalism and election-related civil society organizations, DDoS was the most common," according to Cloudflare. Over the 12 months before the election, the company mitigated over 10.8 billion DDoS requests targeting these groups.

A single attack accounts for a large part of that. On September 10, 2026, a US-based news organization was hit with 3.04 billion requests, the largest single attack in the period, according to the post. If that attack is counted within the 10.8 billion, which the post implies but does not state, it represents about 28% of the year's total (PPC Land calculation). The news organization is not named.

Cloudflare ties the pattern to earlier research. A Project Galileo report released earlier this year found that among all Project Galileo entities in 120 countries, DDoS was by far the most common attack, accounting for 81% of all attacks, according to the post.

php2shell reaches a voting website

The one specific exploit named in the post is php2shell. "On October 21, security researchers disclosed serious vulnerabilities (known as php2shell) in PHP, an open source back-end programming language used by 70% of the world's websites," according to Cloudflare. "The flaws reportedly could allow attackers to execute malicious code on vulnerable servers, expose customer data, or disrupt services." The company says it deployed Web Application Firewall (WAF) protections for all customers before public disclosure.

The disclosure itself was covered on the day. Cloudflare's rules went live at 3:02 am UTC on October 21, about nine hours before public disclosure by Cloudflare's count, and targeted two flaws tracked in the CVE system as CVE-2026-109618 and CVE-2026-109619. PHP released fixes at 13:30 UTC that day in versions 8.5.12, 8.4.27, 8.3.36 and 8.2.35, with CERT-EU rating both flaws 9.2 on the CVSS scale. OpenAI said the same day that Calif had found one of the flaws using GPT-5.6 Sol, with Wiz finding both independently.

On October 22, Cloudflare's PHP defense was triggered more than 9,000 times on a state government election website in the Western US. "This attack was only detected on one election-related site and appears to have been an isolated incident rather than part of a broader trend," according to the company. The post does not name the state, say whether the site ran an affected PHP version, or say whether the 9,000 triggers came from one source.

How Cloudflare's election programs are organized

According to Cloudflare, the protection behind the data runs through three programs, each for a different type of organization, and each offered "at no cost":

  1. Athenian Project - "for state and local election websites."
  2. Cloudflare for Campaigns - "for political campaigns."
  3. Project Galileo - "for civil society, democracy, and journalism organizations."

Together, the post says, they covered more than 1,400 election-related websites during the 2026 midterms, "including those belonging to state and local governments, candidates, political parties, voting-rights advocates, journalists, and civil society groups." Cloudflare launched its first election protection program in 2016 "after widespread reports of cyberattacks targeting the U.S. elections." Eligibility is described on each program's own page; the post itself does not give a breakdown of sites by program, state or type of organization.

Cloudflare's three recommendations

The post closes with recommendations, which are Cloudflare's own and are quoted here in full:

  • "Prepare for surges in user traffic. Legitimate voter interest can create traffic spikes that overwhelm unprepared infrastructure and make essential election websites inaccessible. Organizations should prepare scalable, efficient infrastructure to handle Election Day traffic increases without straining budgets or staff."
  • "Maintain security protections year-round, not just on Election Day. Cyberattacks targeting democratic institutions occur across the entire election cycle, not just around voting windows. Defensive postures must remain active continuously rather than treated as a short-term, seasonal measure."
  • "Deploy automated defenses to counter rapid threats. AI has made it easier for attackers to find and exploit new vulnerabilities. Organizations should implement security services that update automatically, like managed rules, to ensure new protections deploy by default."

The third claim, about AI, comes without supporting data in the post. It matches a line LinkedIn's chief information security officer, Lea Kissner, published on October 30, that the gap between finding a flaw and attempting to exploit it is getting smaller, a claim also made without figures. Managed rules are a Cloudflare product, and the post ends by pointing eligible organizations to its three free programs.

Why the numbers reach beyond election offices

Campaign and candidate websites sit at the end of much of the political advertising bought this year. AdImpact projected total 2026 midterm political advertising at $11.6bn, ahead of the $11.2bn spent in the 2024 presidential cycle, and Magnite told investors on November 4 that it now expected political contribution of over $40 million for 2026, against the $15 million it had projected. The one significant DDoS attack Cloudflare reports, on November 2, fell inside the window in which Meta barred new US political and issue ads, from October 27 through November 3, when only ads already delivering could keep running on Meta's platforms.

Platforms set their own election rules in the same weeks. Bluesky said on October 19 that it would remove false information about when, where or how to vote during the midterms and label unverified claims. Cloudflare's data covers a different layer: the official sites that publish where to vote and who won.

For publishers, the September 10 attack stands out. A single news organization took 3.04 billion requests in one attack, and journalism and civil society groups, not government election offices, were where DDoS remained the dominant threat.

What the post does not say

The post names no state, county, candidate or news organization. It does not define "significant" for DDoS, give the unit behind the "180 times" and "ten times" ratios, state the exact dates of the 12-month window, or count unique visitors rather than requests. It says nothing about whether any of the 392 million vulnerability attacks succeeded, and it offers no data for its statement that AI has made exploitation easier.

Key facts

  • Source document: "Internet traffic and attacks during the 2026 U.S. midterm elections," Cloudflare Blog, dated November 19, 2026
  • Authors: Patrick Day, Jocelyn Woolbright and Carlos Azevedo
  • Source URL: https://blog.cloudflare.com/internet-traffic-2026-us-election/
  • Election Day: Tuesday, November 3, 2026
  • Programs: Athenian Project (state and local election websites), Cloudflare for Campaigns (political campaigns), Project Galileo (civil society, democracy and journalism organizations)
  • Sites covered: more than 1,400 election-related websites
  • First Cloudflare election protection program: 2016
  • Measurement windows in the post: Election Day compared with November 1 (traffic); 12 months before the election (vulnerability attacks, DDoS against journalism and civil society); 90 days before the election (email)
  • Vulnerability named: php2shell in PHP, disclosed October 21, 2026; PPC Land's October 21 coverage identifies it as CVE-2026-109618 and CVE-2026-109619 (CVE numbers do not appear in the election post)
  • Comparison cited: Moldovan Election Commission, Election Day 2025
  • Related Cloudflare report linked in the post: Project Galileo report on cyberattacks against civil society
  • Blog tags: Athenian Project, Cloudflare for Campaigns, DDoS, Project Galileo, Security, WAF

Timeline

Summary

Who: Cloudflare, through its Athenian Project, Cloudflare for Campaigns and Project Galileo programs, which covered more than 1,400 election-related websites belonging to state and local governments, candidates, parties, voting-rights groups, journalists and civil society organizations. The post is written by Patrick Day, Jocelyn Woolbright and Carlos Azevedo.

What: An analysis of traffic and attacks around the 2026 US midterms. Government election websites faced website vulnerability attempts 180 times more often than DDoS attacks, with exploit attempts rising from 200,000 a day in January to 2 million a day in October and 392 million mitigated over 12 months. Election Day brought about 602 million requests to state and local election websites and no significant DDoS attack; the only one reported hit a California House candidate on November 2. Journalism and civil society groups took over 10.8 billion DDoS requests, and one state election website triggered Cloudflare's PHP defense more than 9,000 times on October 22.

When: Cloudflare published the analysis today, November 19, 2026, covering the year to Election Day on November 3, 2026.

Where: The United States, with a comparison to Moldova's 2025 election. The PHP incident involved a state government election website in the Western US.

Why: Election websites carry where-to-vote and results information, and the data shows the threat to them moving from disruption toward attempts to reach internal systems and data, building for months before the vote. For campaigns and publishers, whose sites sit at the end of a midterm advertising cycle AdImpact projected at $11.6bn, the split matters: government sites drew break-in attempts, while newsrooms and civil society groups absorbed most of the flood attacks.