Why Do Bad Actors Compromise Sites

Why would anybody want to hack your website?

It is a perfectly reasonable question. You might run a small business site, a local community newsletter or a blog documenting your cat’s increasingly unreasonable demands.

You aren’t holding state secrets. You probably aren’t holding much money either.

Unfortunately, the person compromising your site doesn’t necessarily care what you publish. Your website has things they can use: a domain, visitors, storage, processing power and connections to other services.

You are paying for infrastructure. They would quite like to use it.

Money is one of the clearest motivations, but there are several ways to turn access to a website into something profitable.

Take phishing. An attacker can place a fake login page on your domain and direct people towards it. The page might pretend to belong to a bank, an email provider or another business.

Your actual homepage could remain completely untouched. You carry on publishing pictures of Mittens while a directory elsewhere on the site is collecting somebody else’s passwords.

The same principle applies to malicious downloads and redirects. An attacker can use your site to send visitors towards scams or malware. The site becomes part of the delivery mechanism, even if neither you nor your visitors were the original target.

Then there is SEO spam.

A compromised site can acquire pages advertising products you have never sold, links you never added or redirects which only appear under particular conditions. Someone is trying to benefit from your domain and its visibility in search results.

The person buying those links may not even know where they came from. Somewhere along the chain, access to your website has become part of a service being sold. Sucuri’s investigations document both SEO spam and malicious redirects on compromised sites. Sucuri’s hacked-website report

This is one reason “the homepage looks fine” is not much of a security check. Keeping the legitimate site working can help an attacker remain unnoticed.

For an online shop, the checkout provides another opportunity.

Malicious code can be added to collect payment information or deceive customers into entering it into an attacker-controlled form. This is usually described as card skimming.

The details depend on how payments are integrated, but the important point is that an attacker may target information while a customer is entering it. You don’t have to be deliberately storing card numbers in your database for a compromised checkout to cause harm. A documented WooCommerce skimmer investigation

Other attackers may try to make money directly from you.

They might damage or encrypt files, interfere with access, steal information or demand payment under threat of further disruption. A ransom demand is a claim made by an attacker; it does not, by itself, establish what they have actually done.

Either way, a site which is important to your business gives someone a potential means of applying pressure.

Your server’s resources also have value independently of the website.

Depending on the access obtained, an attacker may use the hosting environment to send spam, relay traffic, host malicious files or participate in attacks against other systems.

Cryptocurrency mining is another example. You provide the computing resources and pay the bill; someone else receives any proceeds.

The return from one small server might be unimpressive. Repeating the same approach across many compromised systems changes the economics.

AI services have introduced another version of resource theft. Exposed API keys or stolen cloud credentials can give an attacker access to paid model services. That access can be used or resold while the legitimate account holder receives the charges.

This is often called LLMjacking. Recent investigations describe attackers stealing access to hosted models and cloud resources, rather than needing to build their own infrastructure. Google Threat Intelligence Group’s LLMjacking research

For a WordPress site connected to an AI service, the practical lesson is familiar: credentials stored or used by the site are assets too.

Then there is the information you hold.

Customer details, form submissions, private documents and integration credentials may all be useful to an attacker. What is valuable depends on the site and the access they obtain.

Don’t assume that because the public pages are uninteresting, everything behind them is equally uninteresting. Equally, don’t assume that every compromise involved someone carefully reading your database. You need evidence to establish what happened.

Some attacks are about disruption, publicity or a grudge.

A defaced homepage might carry a political message, an attacker’s name or something deliberately embarrassing. A competitor, former employee or campaign group may have a particular reason to target an organisation.

These possibilities matter, especially where the organisation’s work makes targeted attention plausible. But an attacker does not need a personal interest in your business to compromise its website.

Automation makes opportunistic attacks practical.

A script can look for sites running vulnerable software, try a known technique and record which attempts worked. It does not need to understand your business, read your About page or develop strong feelings about Mittens.

That is why being small or obscure provides little reassurance. If your site is reachable and has a weakness the attacker’s tooling can exploit, it may be useful to them.

An automated attack is still an attack carried out by people. Automation changes how they operate and how many systems they can reach.

There can also be several people involved at different stages.

Imagine someone gains access to a shop and sells that access to another group. The buyer installs a skimmer. Stolen information is then passed to others who specialise in using or selling it.

That is an illustrative chain, rather than a description of every compromise, but it explains why the person who gets in may not be the person who ultimately profits.

Access itself is something which can be traded.

This also explains why removing the most visible symptom may leave the underlying compromise unresolved. Deleting a spam page does not necessarily remove the mechanism which created it, or another way the attacker has arranged to return.

Understanding the motive helps you think about what needs protecting.

Look beyond your published content. Consider the people visiting the site, the accounts which control it, the information it handles and the services it can access.

Your website does not need to be famous to be useful to somebody else.

Mittens may be of limited interest to organised crime. The hosting account paying for his online presence is another matter.

The lesson has the following resources:

Sucuri’s 2023 hacked-website and malware report – Examples of SEO spam, redirects and other infections observed in Sucuri’s investigations and scans. Its findings describe that dataset, rather than every compromised website.

WooCommerce card skimmer and backdoor investigation – A documented example of checkout theft alongside mechanisms for maintaining access.

Google Threat Intelligence Group: LLMjacking – Research into theft and resale of access to cloud and AI model resources. This supports the lesson’s AI resource-theft example; it does not establish distributed model training on ordinary WordPress servers.

Update Management (coming soon)

Browser Security (coming soon)

Configuration Management (coming soon)

Hosting and Backups (coming soon)

Monitoring (coming soon)

Compliance (coming soon)

Putting it all together (coming soon)