Summary: Cloudflare’s 2026 U.S. midterm traffic report, published November 19, 2026, covers more than 1,400 election-related websites it protected for free through the Athenian Project, Cloudflare for Campaigns, and Project Galileo. The headline finding inverts the usual election-security story: on Election Day itself there was essentially nothing, while vulnerability-scanning attacks built all year and peaked in late October. For anyone running a small public-facing site with a scheduled big day, the timing of the threat matters more than its size.
Election Day was the quiet part
Voter traffic behaved as expected. Requests to state and local government election websites hit roughly 602 million on November 3, up from 130 million on November 1 — a 363% surge, and the highest day of 2026 for those sites. Traffic across the rest of Cloudflare’s network rose just 12% over the same window, so the spike was specific to election infrastructure.
Attacks, meanwhile, stayed home. Cloudflare recorded no significant DDoS attack against any of its 1,400 protected election sites on Election Day. The only notable one came a day earlier, on November 2: 520,000 requests against a U.S. House candidate in California — small by DDoS standards. The contrast with 2025 is stark; Cloudflare notes Moldova’s election commission absorbed over 898 million malicious requests in a single 12-hour period that year, peaking at 324,333 requests per second.
The real attack was code probing, and it peaked in October
Across the 12 months before the vote, Cloudflare says it mitigated 392 million website vulnerability attacks on government election sites alone. On those sites, vulnerability attempts ran 180 times more often than DDoS, and at roughly ten times the rate seen across Cloudflare’s broader customer base. These are the unglamorous exploits — cross-site scripting to steal admin credentials or deface pages, SQL injection to reach databases — aimed at getting into internal systems and data rather than knocking a site offline.
The volume climbed steadily: about 200,000 attempts per day in January, roughly 2 million per day by October, peaking several weeks before the election.
A concrete example shows how fast the window opens. Researchers disclosed the php2shell vulnerabilities in PHP — which powers 70% of the world’s websites — on October 21. Cloudflare had already deployed WAF protections for all customers before the public disclosure, and on October 22 its PHP defense fired more than 9,000 times on one state government election website in the Western U.S. Cloudflare describes that as an isolated hit rather than a trend. The mechanism — managed rules deployed ahead of disclosure — is one we’ve watched before with Cloudflare WAF rules for the Java4Shell vulnerability.
Email looked normal and wasn’t quite
Phishing volume against political campaigns tracked the ordinary baseline — 3.9% of email to Cloudflare for Campaigns participants was malicious over the final 90 days, against a 4.8% baseline for other Cloudflare customers. The worrying part is bypass: nearly 42% of those malicious emails got past basic email authentication checks such as sender-identity verification and in-transit tampering prevention. Only Cloudflare’s advanced phishing detection caught them.
The mix also depends on who you are. For journalism and election-related civil society groups, DDoS was the top attack: over 12 months, Cloudflare mitigated more than 10.8 billion DDoS requests against these organizations, with the largest single hit — 3.04 billion requests — landing on a U.S.-based news organization on September 10, 2026. That matches earlier Cloudflare research finding DDoS at 81% of attacks among Project Galileo entities across 120 countries.
Our take: If you run a small public-facing site — civic, campaign-adjacent, news, or anything with a scheduled traffic day — act now, and act on the boring parts: turn on managed, auto-updating WAF rules and tighten email authentication this week. The data supports the timing argument Cloudflare itself makes: attacks peaked weeks before the event, pre-disclosure WAF rules caught 9,000+ exploit hits the day after the PHP disclosure, and basic email auth failed 42% of the time on malicious campaign mail. These are cheap, reversible changes that don’t need a platform engineer. One assumption to name: every number here is Cloudflare measuring traffic on Cloudflare-protected sites, and the sources say nothing about what these attempts actually achieved, so treat this as a threat-mix signal, not an outcome audit. Teams whose sites have no scheduled-traffic moment, no admin surfaces or PHP backend, and already run managed rules can ignore this. If independent telemetry showed a different attack mix off Cloudflare’s network, or if managed rules generated enough false positives to break legitimate traffic, we’d revisit.
