Uber Data Breach Exposes Driver License Data via Third-Party AI Vendor

Ride-hailing giant Uber has started notifying nearly 150,000 drivers that their personal information and driver’s license numbers were compromised during an October security breach. Interestingly, the incident didn’t happen inside Uber’s own primary systems. Instead, the attackers broke into a third-party technology vendor, shining a bright light on the hidden vulnerabilities lurking inside modern enterprise supply chains and outsourced cloud services.
According to official filings submitted to state attorneys general in places like California, Iowa, and Oregon, Uber learned on October 9 that unauthorized actors had breached the vendor’s environment. The hackers stole internal credentials, which they then used to slip past security barriers and access data belonging to Uber. It marks the second known cybersecurity incident for the transportation company this year, following an earlier security event that hit its Uber Freight division.
📌 Key Takeaways
- Uber is alerting 149,346 drivers that their personal details were stolen in a third-party data breach.
- The breach took place at Fireworks AI, a developer platform used to access open-weight artificial intelligence models.
- Compromised details include drivers’ full names, home addresses, dates of birth, and driver’s license numbers.
- Attackers managed to bypass detection tools by using valid, stolen internal vendor credentials rather than brute-forcing their way in.
How the Breach Unfolded at Fireworks AI
The security lapse traces back to Fireworks AI, a platform that gives developers the tools to deploy large language models and open-weight AI frameworks. Fireworks reported last month that malicious actors had gained unauthorized access to customer systems, making off with billing records and private datasets. Other prominent tech companies, including productivity startup Notion and AI coding platform Cursor, also confirmed that their data was exposed during the exact same intrusion.
Filings with state regulators show that 149,346 individuals were impacted on the Uber platform alone. Uber stated that it maintains automated alerts to spot unauthorized activity within its Fireworks environment. However, because the intruders utilized legitimate internal credentials belonging to the vendor, those defense mechanisms failed to raise any red flags. A representative for Fireworks attributed the campaign to a financially motivated cybercriminal group hunting for tokens to access large language models, though no specific threat actor has stepped forward to claim responsibility.
The Technical Mechanics Behind the Bypass
To understand how the intrusion succeeded without tripping standard alarms, it helps to look at how modern machine learning cloud integrations work. Fireworks AI operates as a managed infrastructure provider, running open-weight large language models for enterprise clients. This setup requires maintaining persistent API endpoints, database connectors, and administrative service accounts that link external vendor systems directly to client repositories.
The attackers exploited this architecture by launching a credential-stuffing and session-hijacking campaign directly against Fireworks AI’s internal administrative tier. Once they got their hands on valid JSON Web Tokens (JWTs) and API keys belonging to vendor staff, they used those authenticated sessions to query shared storage buckets. Because the requests came from recognized internal IP ranges and presented cryptographically valid tokens, Uber’s automated boundary defenses treated the traffic as normal administrative work rather than an active breach.
This exposes a systemic weakness in how companies handle zero-trust segmentation for third-party machine learning pipelines. Standard perimeter security often assumes that once an identity is authenticated, everything it does next is legitimate. Cybersecurity experts are now pushing hard for behavioral micro-segmentation, which requires continuous contextual checks—such as hardware-backed device attestation—even for trusted vendor service accounts.
Real-World Risks for Impacted Drivers
For everyday drivers, having a driver’s license number leaked alongside a name, date of birth, and physical address carries long-term risks. Unlike credit cards, which can be canceled and reissued in minutes, driver’s licenses are static identity documents. Compromised license data can easily fuel sophisticated phishing campaigns, identity theft, and fraudulent account creations.
This incident also piles pressure on gig-economy platforms to clean up their vendor risk management practices. Rideshare operators handle massive volumes of sensitive regulatory documentation to comply with state transport laws. When those repositories are outsourced to specialized AI or cloud startups, customers and contractors bear the privacy risks. As businesses race to adopt generative AI tools, ensuring uniform security hygiene across every link in the supply chain remains an urgent industry challenge.
Frequently Asked Questions
What specific information was stolen in the Uber breach?
Hackers accessed affected drivers’ full names, residential addresses, dates of birth, and driver’s license numbers that were stored with the third-party provider.
How did the hackers access Uber’s data without triggering alerts?
The attackers breached Fireworks AI and stole valid internal credentials. Because they used legitimate internal tokens to enter the system, Uber’s automated security monitors did not recognize the activity as unauthorized.
Am I affected if I am an Uber driver?
Uber is directly notifying the 149,346 drivers whose data was confirmed stolen. If you have not received an official notice from Uber or your state’s attorney general, your data was not part of this specific disclosure.
What steps should affected Uber drivers take to protect their identities?
Drivers who receive official notification letters should immediately place a security freeze or fraud alert on their credit reports with major bureaus. Because driver’s license numbers are static documents, affected individuals should also monitor state department of motor vehicles records for unauthorized activity and watch out for targeted phishing attempts.
Sources and further reading
Read the original reporting on TechCrunch.