Ten simulated bots walked past 65.3% of the popular sites DataDome tested
DataDome measured malicious bot activity up 124% in the year to June 2026 against 13.2% growth for humans, and its own detection test found most popular sites challenged none of ten simulated bots.
The Watch · Security desk
What happened
- DataDome puts malicious bot activity up 124% between July 2025 and June 2026, measured across trillions of requests to more than 75,000 customer sites, against 13.2% growth in human traffic.
- Bots and AI agents generated about 26.5% of all traffic in that dataset, with traffic from AI agents and large language model crawlers up 82.3% over the same 12 months.
- In a separate test against popular websites, 65.3% failed to detect any of 10 simulated bots and 2.4% stopped or challenged every type.
- Monthly AI bot traffic to login pages rose more than eightfold in the first half of 2026, driven by assistants checking orders, retrieving account information and supporting purchases.
Why it matters
- Who's exposed exposure Login pages now receive authorized assistants and credential testers in the same request stream, so a site that decides on the first request will refuse customers along with attackers.
- What it limits constraint API, mobile and server-side endpoints never run client-side code, so coverage there means inspecting sequence and volume in the request path. That is a different build from adding a JavaScript challenge.
- What it costs cost More than a quarter of requests being automated puts bot load inside bandwidth, origin capacity and rate-limit budgets whether or not anyone is accounting for it separately.
Next story · Security · Oct 7
FortiBleed keeps scanning exposed Fortinet firewalls as toll grows to 400,000-450,000 devicesConfirmed10 publishers
The test set covered simple scripts, bots presenting themselves as recognized AI services, tools with forged browser fingerprints, and automated instances of real browsers, and the requests were sent through residential internet addresses in the United States, Canada and France [23]. Subtracting the sites that challenged everything and the sites that challenged nothing leaves 32.3% with partial coverage [16].
On growth, 124% against 13.2% is roughly 9.4 times the human rate [26]. The 82.3% rise in AI agent and crawler traffic is about 6.2 times it [27]. Human traffic's share of requests fell over the period, and bad bots accounted for most of that change [15].
Scraping was the dominant activity, at 70.9% of bad bot traffic [9]. In a 30-day sample, simple bots were the largest single share, cheap to run and still effective against sites with limited protection, used for scanning, reconnaissance and attacks on easier targets [17]. Average-complexity bots built on browser automation frameworks turn up in scraping, fake-account creation and credential testing [18]. Advanced operators keep their better tooling separate from the basic bots so that a simple automation signal does not expose a campaign [19]. Credential stuffing was flat year on year, with attackers running high-volume attacks, withdrawing to refresh credential lists and rotate infrastructure, then returning at full scale [8][10].
Browser-side signals do not cover this. They can be suppressed or imitated, and requests aimed at APIs, mobile endpoints or server-side services may never execute client-side code at all [20]. User-agent allowlists fail the other way: malicious software can copy the string of a recognized AI service and arrive as a trusted crawler, while an unfamiliar bot misses the rules written for known services [22]. That pushes detection onto the sequence, speed, volume and context of each interaction, because one request may not reveal a visitor's purpose [21].
"Organizations are being asked to make more nuanced decisions with defenses that are still largely built around binary choices," said Jerome Segura, VP of Threat Research at DataDome [12]. "The ability to distinguish a legitimate AI assistant from a credential-testing bot or an automated account-abuse campaign will be critical to protecting customers without blocking beneficial activity," he said [13].
The two halves of the report measure different populations. Traffic share and growth come from trillions of requests to DataDome's own customer sites [6]. The failure rate comes from tests against popular websites, and the report does not state how many sites were in that set [28]. Volume gave little indication of protection: high-traffic sites performed at a similar level to sites with smaller audiences [24]. DataDome attributes the gaps at large organizations to complex infrastructure, disconnected tools and legacy systems [25].
What to watch
- Whether credential stuffing breaks out of its flat annual line in the second half of 2026 after the surge-and-withdraw cycle DataDome describes.
- An independent, non-vendor measurement of automated traffic share that tests the 26.5% figure.
- Whether sites publish agent policies that separate an authorized assistant at a login page from a credential-testing bot instead of blocking both.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence45
- Adoption
- Not scored
- Hype gap+20
- Incentives75
- Confidence55
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Malicious bot activity increased 124% between July 2025 and June 2026.
- [2]
Human traffic grew 13.2% between July 2025 and June 2026.
- [3]
Traffic from AI agents and large language model crawlers rose 82.3% over the same 12-month period.
- [4]
About 65.3% of the popular websites tested failed to detect any of the 10 simulated bots, and only 2.4% stopped or challenged every type.
- [5]
The figures come from DataDome's State of Bot & Agent Security Report 2026.
- [6]
DataDome analyzed trillions of requests across more than 75,000 customer sites during the 12-month period.
- [7]
AI traffic grew more than six times faster than human traffic.
- [8]
Scraping, scalping, spamming and DDoS activity increased during the study period, while credential stuffing remained flat on an annual basis with sharp drops and resurgences throughout the year.
- [9]
Web scraping was the dominant attack type, accounting for 70.9% of bad bot traffic.
- [10]
Credential-stuffing campaigns followed a recurring pattern: attackers launched high-volume attacks, withdrew to refresh credential lists and rotate infrastructure, then returned at full scale.
- [11]
Monthly AI bot traffic to login pages increased more than eightfold during the first half of 2026.
- [12]
"Organizations are being asked to make more nuanced decisions with defenses that are still largely built around binary choices."
- [13]
"The ability to distinguish a legitimate AI assistant from a credential-testing bot or an automated account-abuse campaign will be critical to protecting customers without blocking beneficial activity."
- [14]
Bots and AI agents generated approximately 26.5% of all traffic in the dataset.
- [15]
Human traffic's share of requests declined, and bad bots accounted for most of the change.
- [16]
32.3% of the popular websites tested detected at least one of the 10 simulated bots but not all of them.
- [17]
In a 30-day sample of bad bot traffic, simple bots represented the largest share; they require little effort to operate, remain effective against websites with limited bot protection, and are used for scanning, reconnaissance and attacks on easier targets.
- [18]
Average-complexity bots are common in scraping, fake-account creation and credential-testing campaigns, and use browser automation frameworks that reproduce parts of a legitimate browsing session.
- [19]
Advanced operators tend to keep sophisticated tools separate from basic bots to avoid exposing a campaign through simple automation signals.
- [20]
Browser-side signals cannot provide complete coverage: attackers can suppress or imitate them, and requests targeting APIs, mobile endpoints or server-side services may never execute client-side code.
- [21]
A single request may not reveal a visitor's purpose; an authorized assistant checking an order can initially resemble a bot attempting to access a customer account, so security systems need to examine the sequence, speed, volume and context of each interaction.
- [22]
Malicious software can copy the user-agent string of a recognized AI service and present itself as a trusted crawler, and an unfamiliar bot can avoid rules created for known services.
- [23]
The tests covered simple scripts, bots posing as recognized AI services, tools with forged browser fingerprints and automated versions of real browsers, with requests sent through residential internet addresses in the United States, Canada and France.
- [24]
Traffic volume offered little indication of protection: high-traffic websites performed at a similar level to sites with smaller audiences.
- [25]
Large organizations may have substantial security budgets, yet complex infrastructure, disconnected tools and legacy systems can leave gaps.
- [26]
Malicious bot growth of 124% is about 9.4 times the 13.2% growth in human traffic.
- [27]
AI agent and crawler growth of 82.3% is about 6.2 times the 13.2% growth in human traffic.
- [28]
The report gives the detection test results as percentages of popular websites tested without stating how many sites were in the test set.
Sources
2 independent publishers whose own reporting we read for this story.
- helpnetsecurity.comNearly two-thirds of tested websites fail every bot test
1 article · September 22, 2026
- scworld.comMalicious bot traffic surges 124%, overwhelming website defenses
1 article · September 23, 2026
Topics and entities
Topics
- Web Scraping
- Bot Management
- Credential stuffing
- AI agents and crawlers
Entities
- Jerome Segura
- DataDome
- State of Bot & Agent Security Report