Here’s the timeline Uber gave regulators, in plain terms:
- October 7: someone outside the company got hold of the vendor’s internal credentials and used them to log in to Uber’s account with that vendor.
- October 9: the vendor spotted the activity and warned Uber.
- October 13: Uber’s investigation confirmed the intruder had accessed Uber data held with the vendor, including personal details.
- November 10: Uber started mailing letters to affected people.
An Uber spokesperson confirmed the letters relate to a security incident at Fireworks.ai, which had access to some Uber data.
Fireworks has had a rough autumn. It disclosed one incident in early October, when attackers got into environment variables holding API keys and tokens (the digital passes software uses to talk to other services) for a limited number of accounts. Fireworks linked that attack to a financially motivated criminal group that sells stolen access to large language models, a practice known as LLMjacking. Then, on October 23, a third-party cloud provider hosting Fireworks research machines was hacked, and the next day the intruders used it to run unauthorized work on those machines. That second break-in exposed customer billing details, some datasets and partial model weights.
Uber’s own dates point to the first of those two incidents, the one built on stolen internal credentials. Fireworks confirmed it found unauthorized access affecting customer data but didn’t answer detailed questions. No criminal group has publicly claimed the theft.