input stringlengths 1 785 | output dict | schema listlengths 0 0 |
|---|---|---|
In each instance, the attacks were carried out by targeting these Defense contractors directly via real-time ad bidding. | {
"entities": {
"entity": [
{
"text": "the attacks",
"start": 18,
"end": 29
},
{
"text": "targeting these Defense contractors directly via real-time ad bidding",
"start": 50,
"end": 119
}
],
"action": [
{
"text": "were c... | [] |
Once targeted, an end user only needed to browse to any website, anywhere in the world, which contained a DoubleClick ad-partner embedded window. | {
"entities": {}
} | [] |
Invincea stopped these attacks on the endpoints by containing the delivered exploits in secure virtual containers, while producing the forensics that led to this discovery. | {
"entities": {}
} | [] |
Next we go in some detail about example attacks perpetrated against the defense firms. | {
"entities": {}
} | [] |
It is important to note that the websites we show next that served up targeted malvertising were victims of malvertising campaigns with no knowledge of the malicious ads they were serving up. | {
"entities": {}
} | [] |
These malicious ads were served up by 3rd party networks, who are unwittingly sourcing malicious content. | {
"entities": {}
} | [] |
As we will discuss later, the 3rd party ad networks themselves are falling victim to malicious content campaigns. | {
"entities": {}
} | [] |
A user visited his online fantasy football league homepage at Fleaflicker.com. | {
"entities": {}
} | [] |
As soon as the page loaded, a malicious ad delivered a backdoor Trojan via a Java-based exploit. | {
"entities": {
"entity": [
{
"text": "a malicious ad",
"start": 28,
"end": 42
},
{
"text": "a backdoor Trojan",
"start": 53,
"end": 70
},
{
"text": "a Java-based exploit",
"start": 75,
"end": 95
}
],
... | [] |
Figure 1 shows a screenshot of the page that was visited. | {
"entities": {}
} | [] |
You will notice the two inline ad placements for DoubleClick ad delivery. | {
"entities": {}
} | [] |
The malware delivered came from a compromised Polish website, and would have installed a generic backdoor Trojan. | {
"entities": {
"entity": [
{
"text": "The malware",
"start": 0,
"end": 11
},
{
"text": "a generic backdoor Trojan",
"start": 87,
"end": 112
}
],
"action": [
{
"text": "installed",
"start": 77,
"end": 86
... | [] |
Note the prominent ad placements by AdChoice, a DoubleClick affiliate. | {
"entities": {}
} | [] |
Figure 2 shows an event tree of the exploit and malware delivered from an ad by visiting Fleaflicker.com. | {
"entities": {
"entity": [
{
"text": "the exploit and malware",
"start": 32,
"end": 55
},
{
"text": "an ad",
"start": 71,
"end": 76
},
{
"text": "visiting Fleaflicker.com",
"start": 80,
"end": 104
}
],... | [] |
The event tree in Figure 2 taken from Invincea's Threat Management Console shows the exploited Java process dropped a file called fvJcrgR0.exe, and that it likely came from Pubmatic, an ad delivery network that allows for real time bidding to deliver ads. | {
"entities": {
"entity": [
{
"text": "the exploited Java process",
"start": 81,
"end": 107
},
{
"text": "a file called fvJcrgR0.exe",
"start": 116,
"end": 142
}
],
"action": [
{
"text": "dropped",
"start": 108,
... | [] |
In this instance, the Pubmatic server redirected to a Web server in Poland that dropped the malware. | {
"entities": {
"entity": [
{
"text": "a Web server in Poland",
"start": 52,
"end": 74
},
{
"text": "the malware",
"start": 88,
"end": 99
}
],
"action": [
{
"text": "dropped",
"start": 80,
"end": 87
... | [] |
The timeline below shows the exact times and URLs visited. | {
"entities": {}
} | [] |
Note the number of re-directs from Fleaflicker.com to different outside properties in Figure 3. | {
"entities": {}
} | [] |
Invincea Threat Management provides a quick way to search for an MD5 hash on third party sites (see Figure 4). | {
"entities": {}
} | [] |
By clicking the VirusTotal link, the analyst will see the following VirusTotal report in Figure 5: From the VirusTotal report in Figure 5, you will see that this malware is a Trojan backdoor that would likely be used to download additional malware or to provide remote persistent access to the attacker. | {
"entities": {
"entity": [
{
"text": "this malware",
"start": 157,
"end": 169
},
{
"text": "a Trojan backdoor",
"start": 173,
"end": 190
},
{
"text": "additional malware",
"start": 229,
"end": 247
},
... | [] |
An employee at a defense contractor visited a free Texas Poker online game. | {
"entities": {}
} | [] |
The Poker site had advertisements on the page, one of which launched a similar attack as seen in before on other websites visited by employees at this firm. | {
"entities": {}
} | [] |
It should be noted that Gpokr.com no longer appears to be serving advertisements from their site. | {
"entities": {}
} | [] |
At the time of the incident, as seen in the logs below, an ad window was previously present. | {
"entities": {}
} | [] |
In the event tree shown in Figure 7, you will see that the winning bid redirected to a direct-to-IP site instead of a site via domain name. | {
"entities": {}
} | [] |
Also, above is the first indication of specific executable DLL files. | {
"entities": {}
} | [] |
Searches for these filenames returned zero results on VirusTotal. | {
"entities": {}
} | [] |
This event on September 14 (Figure 8) shows that delivery.first-impression.com redirected directly to an IP address, not a domain name to deliver its malicious payload. | {
"entities": {
"entity": [
{
"text": "delivery.first-impression.com",
"start": 49,
"end": 78
},
{
"text": "an IP address, not a domain name",
"start": 102,
"end": 134
},
{
"text": "deliver its malicious payload",
"start... | [] |
Note the multiple DLL files written to disk and the spawning of rundll32.exe. | {
"entities": {}
} | [] |
At this point, the Invincea-protected host recognized the unauthorized process and reverted itself to a clean state. | {
"entities": {}
} | [] |
In another incident an employee checked their online Earthlink account. | {
"entities": {}
} | [] |
When they replied to an email, a new ad was loaded on a page that attempted to exploit Java. | {
"entities": {
"entity": [
{
"text": "a new ad",
"start": 31,
"end": 39
},
{
"text": "Java",
"start": 87,
"end": 91
}
],
"action": [
{
"text": "exploit",
"start": 79,
"end": 86
}
]
}
} | [] |
This malvertising was from the same IP address seen in other incidents. | {
"entities": {}
} | [] |
You will notice the inline advertisements on this page in Figure 9. | {
"entities": {}
} | [] |
The event tree in Figure 10 notes that this was likely a spear-phish attack. | {
"entities": {
"entity": [
{
"text": "this",
"start": 39,
"end": 43
},
{
"text": "a spear-phish attack",
"start": 55,
"end": 75
}
],
"action": [
{
"text": "was",
"start": 44,
"end": 47
}
]
}
... | [] |
The timeline will show that when the user replied to an email, the ads on the Earthlink page refreshed, dropping the exploit code via Java. | {
"entities": {
"entity": [
{
"text": "the ads on the Earthlink page",
"start": 63,
"end": 92
},
{
"text": "the exploit code",
"start": 113,
"end": 129
},
{
"text": "Java",
"start": 134,
"end": 138
}
],... | [] |
Note in the timeline in Figure 11, how there was a 7 minute gap between the DoubleClick ad redirect and the delivery.first-impression.com ad. | {
"entities": {}
} | [] |
This is an indication that the page was refreshed or the ad was refreshed on the page. | {
"entities": {}
} | [] |
The same exploit IP address from the Gpokr event is present. | {
"entities": {}
} | [] |
This event is the oldest, happening on September 11. | {
"entities": {}
} | [] |
The three examples above are samples of the more than two dozen micro-targeted attacks we have witnessed and blocked as part of Operation DeathClick since mid-September. | {
"entities": {}
} | [] |
Defense Industrial Base customers witnessed micro-targeted malvertising at a rate six times that of comparable private sector companies with similar defense-in-depth capabilities. | {
"entities": {}
} | [] |
We observed in Operation DeathClick that real-time ad bidding networks are being used by criminal enterprise to target companies with malicious content in order to gain persistent remote access. | {
"entities": {
"entity": [
{
"text": "real-time ad bidding networks",
"start": 41,
"end": 70
},
{
"text": "criminal enterprise",
"start": 89,
"end": 108
},
{
"text": "target companies with malicious content",
"start": 1... | [] |
In these third-party arrangements, the content is frequently not vetted because billions of impressions are rendered in real-time. | {
"entities": {}
} | [] |
Most of the content is legitimate ads. | {
"entities": {}
} | [] |
A small fraction is malicious content linking to landing pages that infect users. | {
"entities": {
"entity": [
{
"text": "malicious content",
"start": 20,
"end": 37
},
{
"text": "landing pages that infect users",
"start": 49,
"end": 80
}
],
"action": [
{
"text": "linking",
"start": 38,
... | [] |
Real-time ad networks are being used, often unwittingly, and some have taken steps to try and combat malicious use of their networks. | {
"entities": {}
} | [] |
The Online Trust Alliance is one such industry group comprised of major software companies and ad networks working together to try and address this problem. | {
"entities": {}
} | [] |
Our goal in this paper is to shed light on the micro-targeting of companies by criminal enterprise employing real-time ad networks, and to aid the industry in collectively addressing this problem. | {
"entities": {}
} | [] |
Real-time ad bidding networks have evolved over the last ten years as a means of micro-targeting customers with advertising content they are more likely to click-on. | {
"entities": {}
} | [] |
From Wikipedia: Real-time bidding (RTB) refers to the means by which ad inventory is bought and sold on a per- impression basis, via programmatic instantaneous auction, similar to financial markets. | {
"entities": {}
} | [] |
[1] With real-time bidding, advertising buyers bid on an impression and, if the bid is won, the buyer's ad is instantly displayed on the publisher's site. | {
"entities": {}
} | [] |
[2] Real-time bidding lets advertisers manage and optimize ads from multiple ad-networks by granting the user access to a multitude of different networks, allowing them to create and launch advertising campaigns, prioritize networks and allocate advertising stock. | {
"entities": {}
} | [] |
Real-time bidding is a dynamic bidding process where each impression is bid for in (near) real time, against a static auction where the impressions are typically bundled in groups of 1,000. | {
"entities": {}
} | [] |
A typical transaction begins with a user visiting a website. | {
"entities": {}
} | [] |
This triggers a bid request that can include various pieces of data such as the user's demographic information, browsing history, location, and the page being loaded. | {
"entities": {}
} | [] |
The request goes from the publisher to an ad exchange, which submits it and the accompanying data to multiple advertisers who automatically submit bids in real time to place their ads. | {
"entities": {}
} | [] |
Advertisers bid on each ad impression as it is served. | {
"entities": {}
} | [] |
The impression goes to the highest bidder and their ad is served on the page.This process is repeated for every ad slot on the page. | {
"entities": {}
} | [] |
Real time bidding transactions typically happen within 100 milliseconds from the moment the ad exchange received the request. | {
"entities": {}
} | [] |
The bidding happens autonomously and advertisers set maximum bids and budgets for an advertising campaign. | {
"entities": {}
} | [] |
The criteria for bidding on particular types of consumers can be very complex, taking into account everything from very detailed behavioral profiles to conversion data. | {
"entities": {}
} | [] |
The following infographic summarizes how advanced adversaries are now micro-targeting companies using malvertising. | {
"entities": {}
} | [] |
The marketplace and auction of ads sounds great for actual ads. | {
"entities": {}
} | [] |
But what if the landing pages that are supposed to be ads are actually malicious PHP pages with embedded malware? The bidding and ad placements work the same, but instead of seeing a flashy ad banner, the highest bidder for the placement serves malware. | {
"entities": {
"entity": [
{
"text": "the highest bidder for the placement",
"start": 201,
"end": 237
},
{
"text": "malware",
"start": 245,
"end": 252
}
],
"action": [
{
"text": "serves",
"start": 238,
"... | [] |
The price to win the bid to push malvertising to any page you happen to visit ranges from 45 to 75 cents per impression. | {
"entities": {}
} | [] |
A malicious advertiser on a network may serve crafted, seemingly normal ads, a majority of the time. | {
"entities": {}
} | [] |
In fact, the ads are often stolen copies from legitimate advertisers. | {
"entities": {}
} | [] |
This establishes the attacker's legitimacy and trust on the ad network. | {
"entities": {}
} | [] |
Of course with real-time ad bidding, he can simply offer up low bids and his content would consistently lose in the marketplace. | {
"entities": {}
} | [] |
But it is very simple to replace the redirection code to switch from a legitimate ad banner to a drop site that hosts an exploit kit, typically based on Java, Flash, Silverlight, or all three. | {
"entities": {
"action": [
{
"text": "replace",
"start": 25,
"end": 32
}
],
"entity": [
{
"text": "the redirection code",
"start": 33,
"end": 53
},
{
"text": "switch from a legitimate ad banner to a drop site that hosts... | [] |
Once the malvertiser detects that he has several infected hosts, he removes the redirection code and goes back to serving standard ad banners. | {
"entities": {}
} | [] |
He then'' burns'' his temporary exploit kit drop site, moving his exploits to another location for a new campaign. | {
"entities": {}
} | [] |
This allows the malicious advertiser to perform hit and run attacks, infect whomever he wants at whatever time he wants, and maintain his presence on the advertising marketplace without drawing undue attention to his activities. | {
"entities": {}
} | [] |
In the sections below, we will provide highlights of the RTB industry, its targeting capabilities, and show how malvertisers have been mis-appropriating RTB networks to deliver malware. | {
"entities": {}
} | [] |
The RTB ad networks provide significant micro-targeting capabilities that have long been used to serve legitimate content to users more likely to click on them. | {
"entities": {}
} | [] |
In the following, we describe these capabilities to show the state of the art in RTB network capabilities. | {
"entities": {}
} | [] |
The quoted material below are direct quotes from Real Time Bidding service providers linked. | {
"entities": {}
} | [] |
Emphasis added by Invincea. | {
"entities": {}
} | [] |
Audience Targeting: Bid on the audiences most valuable to you. | {
"entities": {}
} | [] |
Each impression in the PubMatic auction can be enhanced with first- and third-party data; giving buyers targeting capabilities across display, mobile, tablet and video inventory. | {
"entities": {}
} | [] |
Media buyers can also cookie sync with publisher audiences to incorporate CRM, retargeting and exclusion strategies in their digital advertising. | {
"entities": {}
} | [] |
Buyers have access to proprietary audience segments either directly through Private Marketplace deals or through the open market. | {
"entities": {}
} | [] |
With hundreds of parameters available to you, PubMatic has your best audiences waiting for you. | {
"entities": {}
} | [] |
With PubMatic, buyers are able to access pre-defined vertical or audience packages, seasonal packages, publisher and/or site-specific inventory packages as well as pre-selected publisher packages and pricing available in Private Marketplaces. | {
"entities": {}
} | [] |
'' First-Impression Buy-Side offers the granular targeting, tracking, and reporting needed to help our clients make the most of their spend, along with an expert support team to advise when needed. | {
"entities": {}
} | [] |
By leveraging real time buying, First-Impression Buy-Side gives media buyers the full control to maximize the value of an impression. | {
"entities": {}
} | [] |
'' Could Malvertisers Track Exploits and their cost per impression? Yes. | {
"entities": {}
} | [] |
Many RTB networks provide a control panel to track advertising campaigns in real-time, along with notifications that bids have been won and who exactly was served the malware. | {
"entities": {}
} | [] |
Below is a URL redirection log from First-Impression.com from a winning bid by a malvertiser. | {
"entities": {}
} | [] |
In the URL are parameters such as the type of ad, the type of user-agent string of the ad reader specified (which discloses browser and java versions), whether it is a retargeted ad based off of cookies (this one was not), the price paid, which is 65.4 cents, and the notification to the malvertiser that his malvertisin... | {
"entities": {}
} | [] |
DoubleClick discusses their targeting capabilities in online documentation. | {
"entities": {}
} | [] |
Since they specialize in knowing the location of their ad windows, they market those ad spaces to the actual advertisers and malvertisers, along with targeted demographics about the content pages, the visitors to the sites and more. | {
"entities": {}
} | [] |
To showcase the variety of impression-level data available to buyers, consider the data made available through a connection to DoubleClick Ad Exchange's real-time bidding API. | {
"entities": {}
} | [] |
With ADX, a buyer could consider any of the following data passed from the seller with each impression: • Ad slot parameters: visibility (above or below the fold), size, excluded creative attributes, excluded advertiser URLs, allowed vendor or ad technology. | {
"entities": {}
} | [] |
• Geo parameters: country, region, metro, city. | {
"entities": {}
} | [] |
• Content parameters: site URL, site language, seller network, vertical or category. | {
"entities": {}
} | [] |
• User parameters: browser, operating system, anonymous cookie (hashed), cookie age. | {
"entities": {}
} | [] |
Just like when considering one type of data, by using the anonymous cookie parameter, buyers can consider first-party retargeting or third-party audience data from a data provider. | {
"entities": {}
} | [] |
However, they can go further in the evaluation by looking at more of these parameters. | {
"entities": {}
} | [] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.