OTWopentechwire
Tech Intelligence, Openly Wired
Startups

Uber Drivers' Licence Records Exposed in Third-Party AI Platform Breach

Thousands of personal records leaked through Fireworks AI compromise, marking the ride-hail company's second disclosed incident in 2026

AS
Arjun S. Mehta
AI Correspondent · Bengaluru
Nov 18, 2026
7 min read
Uber Drivers' Licence Records Exposed in Third-Party AI Platform Breach
Credit: Marek Antoni Iwanczuk / Getty Images

Breach Disclosed After October Intrusion

Uber has begun notifying thousands of drivers across multiple US states that their personal information and driver's licence records were accessed by attackers who compromised one of the company's technology vendors. Filings submitted this week to state attorneys general reveal that the ride-hail platform learned of the intrusion on 9 October, when the unnamed provider reported that unauthorised parties had obtained internal credentials and used them to reach Uber's stored data.

The affected vendor is Fireworks AI, a platform that enables developers to deploy and run open large language models. Fireworks disclosed its own security incident last month, acknowledging that intruders had stolen customer information spanning billing records and certain private AI datasets. The compromise has rippled outward: productivity software maker Notion separately confirmed that its data was also taken in the same attack.

Uber's disclosure marks the second time in 2026 that the company has reported a security event. Earlier this year, its Uber Freight logistics arm experienced an incident for which an extortion group claimed responsibility, saying it had exfiltrated company data.

Scale of Exposure Emerges Through State Filings

Breach notification letters filed with regulators show that at least 7,870 individuals across three states had their driver's licence information stolen. Massachusetts reported 2,567 affected residents, South Carolina 3,430, and Rhode Island 1,873. A separate filing in Illinois confirmed that some state residents were impacted but did not specify a count, listing the figure as pending. California drivers also received alerts, though Uber has not disclosed how many.

The tally is expected to climb as additional state notifications are processed. Uber declined to provide an overall figure for the number of drivers affected or the total volume of records accessed.

The stolen records include driver's licence details, which typically encompass full names, addresses, dates of birth, licence numbers, and photographs. For gig-economy workers, these documents serve as both identity verification and the legal authorisation to operate commercially, making their exposure a material risk for fraud and identity theft.

Third-Party Risk Amplified by AI Infrastructure Sprawl

The incident underscores a structural vulnerability that has intensified as companies integrate external AI platforms into their operations. Fireworks AI, like many infrastructure providers in the generative AI stack, sits between application developers and the compute-intensive models they deploy. That intermediary position grants access to customer data, API keys, usage logs, and in some cases proprietary datasets used for fine-tuning or retrieval-augmented generation.

When such a provider is breached, the blast radius extends across its entire customer base. Notion's involvement illustrates the breadth: a workspace collaboration tool and a mobility platform, ostensibly unrelated, share exposure through a common dependency.

Fireworks attributed the intrusion to financially motivated actors whose objective was to steal tokens granting access to large language models. These tokens, which authenticate API calls to hosted LLMs, can be resold on underground markets or used to rack up compute charges on victims' accounts. No cybercriminal group has publicly claimed the theft, and Fireworks has not detailed the technical vector of the compromise.

The company confirmed that it identified unauthorised access to some of its systems and that customer data was affected, but it has not disclosed whether the breach involved misconfigured storage, credential phishing, or exploitation of a software vulnerability.

Notification Lag and Vendor Accountability

Uber learned of the breach on 9 October and began driver notifications in mid-November, a gap of roughly five weeks. US state breach notification laws typically require disclosure within 30 to 90 days of discovery, depending on jurisdiction, though companies often cite investigation time as the reason for delays.

The timeline also raises questions about detection latency at Fireworks. The company disclosed the incident publicly in October, but it has not said when the intrusion occurred or how long attackers maintained access before being discovered. In modern cloud environments, dwell time - the period between initial compromise and detection - can stretch from days to months, during which adversaries may exfiltrate data incrementally or establish persistent backdoors.

Uber's reliance on Fireworks for data storage or processing is not publicly documented in the company's technical disclosures, and the nature of the data relationship remains unclear. The ride-hail platform has invested heavily in machine learning for demand forecasting, route optimisation, and fraud detection, all of which could plausibly involve external AI infrastructure. Whether driver's licence records were actively processed by Fireworks models or simply stored within the vendor's environment has not been clarified.

Regulatory and Operational Implications

The breach arrives as US regulators intensify scrutiny of third-party risk management. The Federal Trade Commission has pursued enforcement actions against companies that failed to adequately vet or monitor vendors, arguing that outsourcing does not absolve a company of responsibility for data protection.

For Uber, the incident compounds an already complex compliance landscape. The company operates under a 2018 consent decree with the FTC following a 2016 breach that exposed 57 million users and drivers, an incident the company initially concealed. That decree mandates a comprehensive information security programme and regular third-party audits. While the current breach involves a vendor rather than Uber's own systems, regulators may examine whether the company's vendor risk assessments met the standard required under the order.

Driver's licence theft also triggers obligations under state motor vehicle codes in some jurisdictions, which treat such records as sensitive personal information subject to specific handling requirements. Massachusetts, for example, imposes strict liability for negligent data security practices, and South Carolina's breach notification statute includes a private right of action, allowing affected individuals to sue.

Broader Pattern in AI Supply Chain Security

The Fireworks breach is part of a pattern. Over the past 18 months, multiple AI infrastructure providers have disclosed security incidents. In early 2025, Hugging Face, a popular model repository, reported unauthorised access to some user tokens. Later that year, a misconfigured cloud storage bucket at an AI training data vendor exposed annotated datasets containing personal information.

The velocity of AI adoption has outpaced the maturation of security practices in the sector. Many AI platforms are young companies scaling rapidly, prioritising feature velocity and customer acquisition over hardened security postures. At the same time, the data flows in AI pipelines are complex: training data, inference logs, embeddings, and fine-tuning datasets all present distinct attack surfaces, and traditional perimeter defences struggle to cover them.

For enterprises integrating these platforms, the challenge is compounded by opacity. Vendor security questionnaires and compliance certifications offer limited visibility into runtime configurations, access controls, or anomaly detection capabilities. Third-party audits, where they exist, are often point-in-time assessments that do not capture the continuous changes inherent in cloud-native environments.

What Affected Drivers Should Do

Individuals whose licence information was stolen face elevated risk of identity fraud. Driver's licences are high-value credentials in underground markets because they satisfy know-your-customer requirements for opening financial accounts, renting vehicles, or applying for credit.

Affected drivers should consider placing a fraud alert or credit freeze with the major credit bureaus - Equifax, Experian, and TransUnion. A fraud alert requires creditors to verify identity before opening new accounts; a freeze blocks access to credit reports entirely, preventing new account creation without the consumer's explicit release.

Monitoring bank and credit card statements for unauthorised transactions is essential, as is reviewing credit reports for unfamiliar inquiries or accounts. Some states allow individuals to request a new driver's licence number if the original has been compromised, though policies vary.

Uber has not indicated whether it will offer identity theft protection services to affected drivers, a step that some companies take following large-scale breaches.

Uber's Response and Forward Posture

In a statement, an Uber spokesperson confirmed that the notifications relate to a security incident at Fireworks AI, which had access to some Uber data. The company directed further questions to the vendor.

The brevity of Uber's public comment contrasts with the detailed disclosures some companies provide after breaches, including root-cause analysis, remediation steps, and commitments to enhanced controls. It is unclear whether Uber has terminated its relationship with Fireworks, implemented additional monitoring of vendor access, or conducted a broader review of third-party data exposures.

The company's second breach disclosure in a single year will likely prompt internal and external scrutiny. Investors, regulators, and drivers may ask whether Uber's vendor risk programme is adequately resourced and whether the company has sufficient visibility into how its data is handled across the supply chain.

For Fireworks, the incident represents a credibility test. AI infrastructure is a competitive market, and customers evaluating platforms weigh not only performance and cost but also trustworthiness. A breach that affects multiple high-profile clients and involves the theft of access tokens - core to the platform's value proposition - may accelerate customer churn or complicate new sales.

Lessons for the Ecosystem

The Uber-Fireworks incident offers several lessons for companies navigating the AI supply chain. First, data minimisation matters: limiting what third parties can access reduces exposure. If driver's licence records were not necessary for the AI workloads Fireworks supported, they should not have been within the vendor's reach.

Second, contractual controls must be paired with technical enforcement. Service-level agreements that mandate encryption, access logging, and incident response are meaningless if the customer cannot verify compliance in real time. Technologies such as confidential computing, which encrypts data in use, and tokenisation, which replaces sensitive data with non-sensitive proxies, can reduce the value of stolen data even if a breach occurs.

Third, incident response planning must account for vendor failures. Companies should pre-define notification workflows, legal obligations, and communication strategies for third-party breaches, rather than improvising under pressure.

Finally, the AI sector must mature its security norms. As these platforms become critical infrastructure for everything from customer service chatbots to autonomous vehicle routing, the consequences of compromise will only grow. Industry consortia, regulatory guidance, and customer pressure all have roles to play in raising the baseline.

Read next
Startups

Bharat Semi Raises $24 Million to Build AI-Powered Image Sensors for Defence and Industrial Markets

Priya Nair · 6 min
Startups

Nvidia's Founder Explores CEO Comeback as Board Faces Pressure Over Ouster

Marcus Halloran · 6 min
Startups

Nvidia's Leadership Crisis Exposes Fragility of AI Investment Ecosystem

Linh T. Pham · 6 min
Spot something wrong? Email corrections@opentechwire.com. We log every correction publicly.