The Oracle-Walmart-TikTok Deal Is Not Enough
The social media deal raises issues involving data custodianship and trusted tech partnerships.
The Oracle-Walmart-TikTok deal has stoked significant interest among security practitioners, and the impact of this partnership is already casting a long shadow of doubt over security best practice.
What was initially anticipated to be a full sale of TikTok's US operations to an American company has morphed into the somewhat less fulfilling "trusted tech partnership" under which Oracle will have "the authority to check the source code of TikTok USA." Unfortunately, that falls well short of full visibility into the source code compilation, deployment, and implementation of any new versions of the TikTok software or its underlying recommendation algorithm. Because a key underlying rationale for this deal was to secure the data of American users of the app, the fact that full visibility into the code is now out-of-scope calls into question who benefits from this seemingly incomplete arrangement.
Which largely explains why I felt compelled to write this piece: To expand upon the concept of data custodianship, best practices for implementation, and why it's critical — especially for a deal that originated due to lack of data trust/security.
Speculation has been rampant around what the Oracle-TikTok deal means, and how it addresses concerns around the privacy of US customer information. I'm but one of countless analysts who's been scrutinizing each release, and I'm troubled by what I've seen. Based on everything I've read so far, the deal comes nowhere near ensuring that US consumer data and national security are safeguarded. In fact, this rather diluted arrangement offers a false sense of security and control that could further imperil end users and their data as it establishes a dangerous precedent for future security-centric deals in the space.
The arrangement falls well short of the originally proposed buyout of ByteDance's American assets by an American company. Instead, Oracle would host US consumer data in its cloud, and Walmart would be the "commercial partner." Oracle and Walmart would own 20% of the newly formed TikTok USA. The software, including the artificial intelligence-driven recommendation algorithm and source code, would remain proprietary and under the ownership of ByteDance. Oracle would, however, have the authority to review the source code.
That last piece causes great concern for me because the process as it currently stands seems to lack chain of custody. And in security circles, that is a huge omission.
"Chain of custody" refers to the chronological or sequential events around the control of information. It's often seen in law enforcement, in the handling of evidence from a crime scene, but it's evident across everyday life, as well. If you've ever ordered a bottle of wine at a restaurant, you've also experienced it: The server brings you the sealed bottle for your inspection, the bottle is opened, and your initial sniff and taste of the poured sample is all done right in front of you. The bottle never leaves your table — or your sight.
False Sense of Security
To illustrate the false sense of security and control that Oracle is facing, let's head into the kitchen. Suppose the chef brings out a cut of meat for you to inspect. You nod your approval, and the chef takes the cut back to the kitchen to prepare for you. A little later, your server brings out your dish, which you accept as being prepared with the cut of meat you inspected, without second thought.
Can Oracle afford to make the same assumption? Can the trusted American partner be absolutely certain that the source code it inspected is the same, unadulterated code that has been compiled and released in a TikTok update? How can it be 100% sure without having full visibility into the actual development process? What about the additional libraries that are referenced in the code? Can Oracle ensure that a function call to an external library is the same that it has reviewed? And what about any incremental code revisions that regularly occur in the course of software development, the minor corrections to fix an issue discovered in the debugging process?
Without this level of visibility and scrutiny, it would be impossible for Oracle to certify that the code can be trusted, and that there aren't any surprises lurking inside it. While I agree Oracle being the US data custodian is certainly a good start, this is nowhere near enough. Oracle needs to know what happens with the data when it is processed by the software, and that data is only going where it is supposed to.
Remember that the original premise for China-based ByteDance's need to divest TikTok was due to the lack of trust that US consumer data and national security was being protected. The Oracle-Walmart-TikTok deal doesn't go far enough to change that.
Simply put, Oracle needs to be in the kitchen with TikTok USA and ByteDance for all of this to be truly effective in safeguarding US customer information and national security. "Trust but verify" means little in a deal originally founded on mistrust, and structured without verification.
For security professionals, the concern around data security is nothing new. If anything, those charged with maintaining security must remain vigilant and ensure that the digital assets of their organization are protected and secure. All social networking applications, regardless of where the company originates or where the data is hosted, present their own serious security concerns. Oracle's trusted partnership status notwithstanding, this deal does little to allay some very pertinent fears.
This is Oracle's first foray into the world of social networking, and its control over processing of the data will be very limited due to the proprietary nature of the technology. Any reassurance that Oracle offers on the security of end-user TikTok data needs to be considered in the same light as with any social networking service. Stay on your guard and ensure that your end users are in the know!
About the Author
You May Also Like
DevSecOps/AWS
Oct 17, 2024Social Engineering: New Tricks, New Threats, New Defenses
Oct 23, 202410 Emerging Vulnerabilities Every Enterprise Should Know
Oct 30, 2024Simplify Data Security with Automation
Oct 31, 2024