Despite the common fear that bots are crawling their way to online domination, it looks like humans are finally taking back the Internet.

According to the latest annual “Bot Traffic Report” from Imperva Incapsula, a cloud-based security service, human traffic overtook bot traffic for the first time in 2015, accounting for 52 percent of all website activity—a jump from roughly 44 percent in 2014.

In fact, human traffic has been steadily increasing since 2013. To be clear, while many bots are nefarious, not all are bad. Traffic from good bots, such as search engine crawlers and marketing research tools, has been steadily declining. Traffic from bad bots, such as hackers and spammers, has remained mostly static.

What’s accounting for this rise in human traffic and overall decline in bot traffic?

Let’s take a look.

Bot traffic to large and more popular sites (100,000+ daily visits) decreased from 54 to 40 percent this past year. With bad bot traffic remaining mostly steady, Incapsula found that good bots were largely responsible for this dip.

The study states, “The more popular a website got, the harder it was for the good bots to keep up with the influx of human and bad bot visits.”

As the report points out, there are two main reasons why good bots might visit a site: indiscriminate crawls (like search engine bots or marketing research tools) and targeted scans to learn more about your website (like uptime or SEO). “In both cases,” the report explains, “high website popularity is unlikely to translate into increased good bot traffic.”

Google cracks down on spam

Since 2012, Google has publicly waged a war against spam. Most notably, the company changed its search algorithm to punish websites that distribute spam. Google also has a secretive antifraud team of more than 100 people dedicated to hunting bots and stopping cybercriminals.

Google’s actions likely influenced a noticeable decline in spam bot traffic, which, according to Incapsula, dipped from two percent in 2012 to 0.1 percent in 2015.

Humans spend more time online

The number of people using the Internet increased by over eight percent last year, jumping from 2.94 billion to 3.17 billion. However, this rise in human traffic doesn’t mean the battle of the bots is over just yet.

Why bots still pose a threat

According to the Association of National Advertisers (ANA) and White Ops, a cyber-security company, ad fraud is set to cost brands $7.2 billion this year, up from $6.3 billion in 2015.

So, if good bot traffic is decreasing and bad bot traffic remains relatively steady, why is fraud on the rise? It turns out there are a couple of key reasons.

Marketers continue to pump more money into their digital ad budgets, meaning there’s more ad spend for bots to rip off. Marketers across the globe are expected to spend a collective $200 billion in digital ad spend this year—a 15 percent increase from 2015.

And not only are marketers spending more, but they’re also buying more ads that are vulnerable to bot fraud. As ANA pointed out, there are two types of ads that are more vulnerable to bots than the average online ad:

1. Video ads: They’re expensive and provide a stronger economic incentive for bot operators to game.

2. Programmatic ads: Marketers have less control over exactly where they’re placed, which make them less secure.

While these types are susceptible to bots, they’re also attractive to marketers. Video spots can be more engaging than articles or images, and programmatic ads can reach wide, targeted audiences with little effort.

Together, this is a dangerous recipe for marketers.

How marketers can stay vigilant

Given this new information, what can marketers do to protect their ads and combat bot fraud? For one, they should be careful about their media buys.

For instance, while there are many advantage to running programmatic placements, marketers shouldn’t attribute the results they see from those ads solely to humans. Businesses with small and medium-sized websites, in particular, should also be cautious of their traffic since those sites may not have the budgets to increase security and monitor bot traffic.

Above all, marketers should invest in creating content that appeals to humans—that is, content that elicits an emotional response and is highly shareable, bumping up viral website traffic on popular sites, which are less likely to be ruled by bots.

It’s a good sign that humans are taking back the Internet, but with ad fraud climbing, publishers and marketers still have a lot of work to do if they want to make sure that the digital ad economy stays as healthy as possible.