Hi, this is Tom Gallagher from the Office Trustworthy Computing team. At Blue Hat v9, David Conger and I presented some of the security engineering work that we were doing to help ensure the security of Office 2010. We don’t want a single bug in our parsing code to allow arbitrary code to harm a customer’s machine by doing things like installing a rootkit. While fuzzing remains an important piece of our security efforts, we’ve taken a layer approach to provide protection against bugs that we may have missed. Key layers of our protection include Office file validation and a sandboxed viewing mode known as Protected View for Word, PowerPoint, and Excel.

After our Blue Hat presentation, we had the privilege of attending and presenting at CanSecWest. For me one of the most interesting parts of security conferences is getting to talk with others about what they are working on. Several people working for both software vendors and security consulting companies were able to talk about our current fuzzing work and challenges over dinner. People and organizations including Microsoft continue to make large investments in finding fuzz bugs. There are a lot of creative ideas that are being explored and will likely continue to be an area of focus for the near future.

The Security Development Lifecycle (SDL) provides a great baseline to help teams ensure we ship secure products. The SDL has evolved since its inception in 2004; the tools and guidance have improved with the help of many product teams across Microsoft. Office is one the contributors of both tools and guidance for the SDL. In Office, we often create additional recommendations and requirements for our code and “dogfood” these ourselves before recommending that these guidelines become part of the official SDL. Last week, Didier from the Microsoft Security Engineering Center made a good blog post about Office’s SDL efforts which outlines many of the specifics.

While we have shipped Office 2010, we know people will continue to attack our code and our efforts to find vulnerabilities haven’t stopped either. Over the past month, we’ve continued to make additional improvements to our Distributed Fuzzing Framework and are now consistently completing over 20 million iterations each weekend. We’re also adding additional fuzzers and tweaks to our fuzzing job scheduler. I’m looking forward to talking more about our latest work at SyScan in Singapore next month and in HangZhou in early July.