← Back to blog
November 14, 2026

Supply Chain Vulnerability in the Age of AI Partnerships

Photo by Erik Mclean on Unsplash

Security is rarely about the front door anymore; it is about the dozen third-party contractors who have their own keys to the side entrance. Uber is relearning this lesson the hard way this week.

The ridesharing giant recently acknowledged that hackers managed to exfiltrate the driver's license information of thousands of its partners. This did not happen because someone guessed a password at Uber HQ. Instead, the breach occurred through Fireworks AI, a cloud company Uber uses for its artificial intelligence workflows. While the number of affected drivers is relatively small compared to Uber’s global scale, the sensitivity of the data—government IDs—makes it a significant headache for everyone involved.

The AI Gold Rush Meets Security Debt

In the rush to integrate AI into every facet of business operations, companies are moving data into new, often less-vetted environments. I have seen this cycle before in the early days of the cloud and again during the move to managed containers. The business team wants the feature yesterday, so the technical team spins up a partnership with an emerging AI platform, and the security team is left trying to audit a moving target.

For the hosting and infrastructure world, this is a cautionary tale about the "AI supply chain." When you outsource your intelligence, you are also outsourcing your liability. If you are a hosting provider or a SaaS company building features on top of third-party LLMs or AI cloud providers, you need to be asking where that training and inference data actually sits and who is watching the logs.

It is somewhat poetic that a company built on disrupting the taxi industry is now being disrupted by the very bleeding-edge tech they used to stay ahead.

The Bottom Line

Uber will survive this, but for smaller players in the ecosystem, a breach through a third-party vendor can be a terminal event for customer trust. We need to stop treating AI as a magic black box and start treating it like any other high-risk database—because that is exactly what the hackers are doing.