A pair of security researchers dominated the annual Pwn2Own contest at the CanSecWest conference in Vancouver last week, driving away with $375,000 in bounties for their six successful hacks—including that of a Tesla Model 3's infotainment system, a feat that earned them the car itself as a prize.
The out-of-balance results are not a fluke. The contest comes only months after vulnerability researchers and security professionals debated the skewed economics of the bug bounty marketplace, where a few researchers win big, hundreds more may make enough to live, and the vast majority often never place.
In a chapter for the book "New Solutions for Cybersecurity," security professionals teamed up with experts from Harvard University and the Massachussetts of Technology to analyze the economics of the bug bounty market. The research, based on data from a Facebook bounty program and 61 programs run by bug-bounty services firm HackerOne, found that the distribution of rewards was in no way equal. A small portion of researchers were big winners, while the vast majority only found a single issue.
"We were looking at the system dynamics of the bug hunting space and the labor market in particular," says Katie Moussouris, one of the authors of the chapter and founder and CEO of Luta Security, a security services firm. "There were very few individuals at the top."
The Pwn2Own competition demonstrated the effect quite well. Team Fluoroacetate showed off six different attacks at the conference. All other teams only accounted for four other hacks, and one of those was only a partial success. (One team withdrew from an attempt to hack the Tesla.)
Without a doubt, the two researchers—Richard Zhu and Amat Cama—deserve to win: This is not the first Pwn2Own competition that Team Fluoroacetate has swept. In November 2018, the duo—whose team name likely comes from a combination of their Twitter handles, RZ_flourescence and Acez—also won $215,000 in the mobile-focused Pwn2Own Tokyo contest. Zhu had previously also won last year's Pwn2Own competition, taking home $120,000.
The Vulnerability Lottery
Bug-bounty programs are a big gamble for researchers, says Moussouris. Researchers invest time and if they are not the first to submit their bug, they will likely not see any reward for their effort. The process rewards those researchers who can find a lot of bugs, fast, she says.
"In bug bounties, if you are not the first one to report it, your work is not compensated—you do not get paid at all," Moussouris says. "Either you are first in, or you wasted your time."
Yet, while the vulnerability-reward economy has only a very thin strata of "middle class" researchers, who make a living wage, the depth of that middle class also depends on where the researcher lives and whether vulnerability research is their only source of income, stresses Dustin Childs, communications manager for Trend Micro's Zero-Day Initiative, which sponsors the competition.
"Maybe you are in a part of the world. where you don't need to make $100,000 a year to live well," he says. "While I know that we do have people who report vulnerabilities to our program who are full-time bug finders, others do it as a side hustle."
The benefits of the programs, however, are real. While the Tesla hack garnered the most attention, some of the best research had a lot less flash, says Childs. Team Fluoroacetate showed a combination of three vulnerabilities that broke out of Microsoft's Edge browser, through the virtual client, and executed code on the host operating system.
"Within in the VMWare client, they opened up an Edge browser, and browsed to a Web page, and that was the only interaction that they had with the machine," Childs says "The potential impact of executing code on the underlying hyper visor is—wow—it was really a great set of bugs."
The exploit chain earned the duo the top prize of the competition: $130,000.
Another contestant used an exploit to break out of Oracle's VirtualBox and attack the underlying operating system.
In addition, the contest demonstrated that researchers are focusing on new areas of code to exploit.
Another contestant, Niklas Baumstark, used a vulnerability in Mozilla Firefox's JIT compiler along with a logic bug to escape that browser's sandbox. And, the hack of the Tesla's infotainment system used a bug in the just-in-time compiler for the user-interface rendering component.
The reliance on JIT to run code inside browsers—and browser-like components—is not surprising, says Trend Micro's Childs.
- Bug Bounty Awards Climb as Software Security Improves
- Crowdsourced vs. Traditional Pen Testing
- White-Hat Bug Bounty Programs Draw Inspiration from the Old West
- Bug Bounty Programs are Growing Up Fast and Paying More
Join Dark Reading LIVE for two cybersecurity summits at Interop 2019. Learn from the industry's most knowledgeable IT security experts. Check out the Interop agenda here.