Could Chat Transcripts Be Security Minefields?
This column was written by Evan Schuman, the editor of StorefrontBacktalk, a site that tracks retail technology, e-commerce and security issues. Retail Realities appears every Friday. Evan can be reached at e-mail and on Twitter.
When Rite-Aid and Walgreens both announced pharmacist chat programs last month, they were the latest chains to try and use chat to get closer to their customers. But, ironically, the preservation of chat discussions of super-sensitive patient medical history may prove a very serious threat to that security.
It's ironic because both chains are taking substantial steps to secure the access to the confidential data, but neither is specifying steps to protect transcripts of that very same data. Imagine forcing call center employees to comply with all PCI rules regarding not preserving prohibited payment
card data, but then allowing them to write all of that data down in plain-text files that are then transmitted to consumers (who are unlikely to protect them) and then saved in the chain's files.
Of the two chains, Rite-Aid opted for the more conservative approach to privacy. Rite-Aid's live chat pharmacists will have no access to patient medical records and will instead only react to what the customer chooses to share during the exchange. But once those customer-shared thoughts are preserved in the chat transcript text file, they can be later accessed.
In the eyes of HIPAA and lawyers for consumers whose data may get accessed, it won't make much of a difference who said the protected data. They will assume that a retail conversation-in this case, a patient-to-pharmacist conversation-will be protected as well as any sensitive medical data.
Privacy is a critical concern for almost all retailers these days, with the few anti-privacy moves-such as Amazon's Patent application to give strangers gift ideas from your purchase history--quite noteworthy. Even Web browser privacy modes are not being found to be especially private.
Walgreens is allowing its pharmacists to access full pharmacy histories for all Walgreens customers, but they're not supposed to reveal anything until the patient has verified identity by answering questions. (Given that Walgreens current pharmacists have always had access
to that kind of nationwide data, it's not a change. That may not be comforting, but it's not a change.) According to Walgreens spokesperson Jim Cohn, the live chat sessions are encrypted. But given that the consumer has to be able to read the answers, it's unclear how secure those communications could be. Even if we assume, however, that they are fully secure, it's unclear how secure the transcripts of those chat sessions will be.
Asked about how the chain handles the chat transcripts, Cohn E-mailed: "Any storage of a chat
session is for internal use only and is stored separate from patients' profiles." That's good, but what security is placed on those records? If they're not part of the patient profile, why are they saved?
Are they excluded from system backups? If an intruder searched for the patient's name, could it be discovered? Or if the system was searched for particular drug names, could it pull up all chat discussions where it was mentioned? How much identifiable data is in those chat transcripts?
As for identifiable data, those chats are also likely to include the answers to those authentication questions. And if those are the keys to unlocking the full patient history, those pieces of text could be among the most dangerous.
There's even a risk on the consumer end. The chain sends the consumer the transcript, including all of the aforementioned secret details. What if that copy gets intercepted? The retailer is in a much stronger position, but will consumers argue that the chain had an obligation to better secure that data? This will be a particularly persuasive argument for elderly patients, who are
the most natural victims for identity thieves and who tend to trust the pharmacies to protect their information.
What about search engine spiders? Are these chat transcripts protected enough to prevent them from accidentally turning up in a generic Google search? ("Hello, law offices of Sue-Em, Sue-Em and Wynne. May I help you?")
This reminds me of the call center issues, where representatives take great care to protect payment card data, but then read the card numbers aloud to verify. What if a dishonest call center rep is recording anything picked near their cubicle? What if a cyberthief is calling the call center repeatedly and recording those calls, trying to pick up the card numbers being read aloud in the background?
All of the security in the world will be made meaningless by the weak link. If not properly handled, chat transcripts of sensitive discussions might be just that.
By Evan Schuman
Special to CBSNews.com