|In the end, poost-market medical device |
security is about peeople and respoonsibility. Photo
taken today outside the Washington Convention
Center meeting on automotive cybersecurity.
I'd like to commend FDA for releasing this difficult to write document. To the arm chair engineer, one might think this is easy stuff. Wrong. While the already finalized pre-market guidance primarily focuses on basic engineering practices to build security into medical device designs, the post-market guidance is mostly about people and effective communication. Why is writing post-market guidance so difficult? Because it's more about people than technology.
There's a lot that the FDA guidance gets right, and most of my criticism pertains to word choice (and lack of puns) that can be solved by editing. The preamble of the document (that focuses on networked medical devices) does not entirely match the body of the document (that talks about all devices, not just networked). The terms "networked devices” and “connected" are red herrings. A network is not necessary for an cybersecurity exploit; malware gets in just fine by unhygienic USB drives carried by unsuspecting personnel. Social engineers still use telephones to trick personnel into enabling unauthorized remote access. The final post-market guidance will need to more deliberately draw attention to outcomes of compromise and risks of vulnerabilities rather than the constantly evolving modality of delivery of exploits. After all, when we talk about surveilling for the spread of flu, we don't limit discussions to spread by cough versus spread by sneeze. Should the document list networked and connected devices as examples of infection vectors? Yes. Should it mention only networked and connected devices? No. Outcomes, not modalities.
What's my opinion on important post-market activities in general? Stakeholders need to communicate vulnerabilities more effectively, and monitor for shifting threats. Medical device manufacturers should create workflows to receive outside input on potential vulnerabilities.
Security folks who discover potential problems need be aware of timescales for responses to responsible security vulnerability disclosures. As Allan Friedman explains, even if you think you're the most important person on the planet, don't expect a medical device manufacturer to simply drop everything they are doing to fix a security flaw overnight. On the other hand, it boggles the mind why a manufacturer might wait a year to meaningfully respond to a clinically significant vulnerability reported by a security researcher.
I expect to see more FDA actions on the less noble manufacturers who do not catch up with this basic medical device security post-market guidance.