There have only been 18 zero day attacks in the past four years, but attackers in possession of these exploits have on average 312 days to wreak havoc, according Symantec Research Labs.
The figure comes from a study by Symantec Research Labs’ researchers Leyla Bilge and Tudor Dumitrasof, which looked at 18 zero day attacks between 2008 and 2011.
Zero day flaws are valuable to an attacker who has a working exploit because targets have few defence options: the vendor whose software it affects has not released a patch while antivirus vendors have not developed a signature for threats associated with it.
Amongst the summary of findings is that “current policies and technologies” do not detect zero-day attacks in a timely manner. The findings were based on telemetry data collected from 11 million Windows hosts that run a Symantec product such as Norton.
Amongst the threats they analysed are well-known malware, including Stuxnet, Ramnit and Conficker. In most cases, the exploits used zero day flaws in Microsoft and Adobe software.
The researchers point out that although most zero day flaws are not used in widespread attacks, three in their sample, including Conficker and Stuxnet, affected hundreds of thousands or millions of hosts.
The risk to organisations is the longevity of the zero day, which may allow an attacker to re-use the same exploits against multiple targets over lengthy periods that in the study ranged from 19 days to 30 months.
The threat to the masses typically comes after the zero day flaw has been disclosed. Symantec detected up to 85,000 new variants each day after the disclosure of each flaw. Its telemetry data also showed that exploits for 42 per cent of the vulnerabilities appeared within 30 days of the disclosure date.
The researchers derived the “when and where” of the attacks by identifying malware on its records that used a flaw before it was disclosed.
“When we find records for the presence of a malicious executable in the wild before the corresponding vulnerability was disclosed, we have identified a zero-day attack,” the researchers explain.
They also claim that 11 of the 18 flaws were not previously known to have been employed in zero day attacks, suggesting zero day attacks are more common than previously thought -- “perhaps twice as many”, the authors speculate.