There are very few articles that I would consider ‘rerunning” in Risky Business however this is one that is worth repeating.
When I read there was going to be a social engineering competition at this year's Defcon (the annual hacker gathering held every summer in Las Vegas) I knew I had to enter. It was the perfect chance to hone my sweet talking skills in a judged and neutral setting, and also to test my hypothesis that not only is social engineering a risk to regular end users in an enterprise, but that even corporate InfoSec teams are not immune from the threat.
Social engineering is essentially "pretexting" yourself into getting people either to divulge sensitive information or getting into areas you otherwise shouldn't be in. You're a sprinkler inspector who shows up unannounced at the front desk or the harried internal auditor racing to meet a deadline who calls an employee seeking information about their computer system. Given most people's inherently trusting and helpful nature, social engineering attacks are surprisingly successful, which is why most corporate information security training programs address the threat. One would assume this would mean the InfoSec groups should be aware of any such attempt. As Gershwin would say, "it ain't necessarily so."
For the Defcon contest, each entrant was randomly assigned a major corporation as a target. Mine just so happened to be the manufacturer of my car. Sweet revenge. The first task was to create a dossier on the target company, solely from information gleaned from the Internet and public sources. There were to be no pre-contest calls, visits to the company's headquarters, or contact with the company in any way whatsoever.
After crawling various search engines for email addresses, phone numbers, addresses, press releases, and other valuable information, I moved onto social networking sites like LinkedIn and Facebook. Soon I had accumulated almost 1,000 email addresses, hundreds of recent press releases, and a couple of employee handbooks as a good starting point. Next, any email addresses not correlated to a name were cut, as were any that couldn't be verified as current or recent employees. The remaining email addresses were then fed into the various search engines to pick out only employees that worked in my target's information security group. These were in turn fed back into the search engines to see if anything interesting fell out; information like hobbies, school affiliation, etc.
Soon the file was whittled down to approximately 75 people that I had gathered at least two points of information with which I could engage them during my pretexting in order to gain an elevated level of trust. If the target person had an interest in flying, I'd be sure to work a local air show into the conversation. Building any sense of familiarity or commonality with a target boosts the success rate of a social engineering attack exponentially.
Next I had to develop the attack vector I would use. Since I was targeting the InfoSec group, I knew I couldn't use the old standby of posing as an auditor for the company. That is such a commonly used ploy that most if not all InfoSec employees should be able to sniff that attack out a mile away. Instead, I settled on posing as a survey taker for CSO Magazine. That would give me cover for calling the security group and asking questions about their security environment.
Three weeks later I was in the soundproof booth at Defcon, dialing through my list of numbers in front of a live audience as I perused the list of "flags" the judges had given me to collect; essentially pieces of information useful for a hacker attack. The first number to answer gave me hope that my hypothesis might be wrong. The security engineer at the other end of the line was very hesitant to speak with me, and very quickly shut me down, refusing to answer any questions that would reveal any technical information about the company. That was a promising sign – perhaps training of InfoSec personnel was starting to become effective.
Not so much. My next target was another security engineer, who, although initially having misgivings about speaking with me, was quickly convinced to participate through both my pleading that I only needed 10 minutes of his time and that I was risking losing my job if I didn't meet my daily target, but more importantly, that there was a $25 iTunes gift card waiting for him upon completion of the "survey". Greed is always a good motivator. Within 15 minutes I had sweet talked the guy into revealing everything from the OS version and service packs installed, browser type and version, to his anti-virus engine and signature version. Basically anything needed to launch a successful targeted attack.
So much for training
In the end I had proven my point; InfoSec people are no different from other end users. While they may have more security awareness training than others, they are still susceptible to the same weaknesses of others; greed, a desire to assist, and a fear of getting in trouble or creating delays in "mission critical" tasks. More important is that they suffer from the same weakness that everyone seems to suffer from – the belief that they would know if they were ever being "snookered."
So what can be done to protect against social engineering attacks? To prevent on-site attacks make this the golden rule that is *never* broken – "unannounced visitors aren't let in if their corporate sponsor isn't reachable to validate the visit." To prevent general social engineering attacks focus your efforts on ongoing awareness training (once a year is not enough), routine testing of personnel to see how effective the training is, and most important of all, reducing information leakage. The amount of information that companies allow employees to post about their jobs and corporate environment is shocking (not to mention the information the companies themselves leak). Take an hour and peruse the various social networking sites liked LinkedIn and Plaxo and see what information you can glean about your employees, the projects they are working on, and what software they are using. Regularly run your company's name through the various search engines to look for information coming from unlikely sources (it's not unusual for contractors or suppliers to post information about dealings with other companies which inadvertently leaks helpful information to an attacker). Doing this exercise from the point of view of an attacker or competitor who knows nothing about your company will allow you to quickly see how many pieces of seemingly disparate information can eventually form a cookbook for a successful attack.
Train and monitor your staff, plug the leaks, monitor the web. Take these three steps and you will be on your way towards reducing (but never eliminating) the threat of social engineering attacks.
Shane MacDougall is a principal partner in Strategic Intelligence, a Canadian-based corporate intelligence gathering firm. He has been a professional white hat hacker, security consultant, and speaker since 1989.
