Skip to content

Uber data breach exposes thousands of drivers’ license records

Uber data breach exposes thousands of drivers' license records

Uber is notifying thousands of drivers after hackers accessed their personal information through compromised credentials belonging to third-party AI infrastructure provider Fireworks AI.

The ride-hailing company disclosed the incident in November 2026 filings with US state attorneys general. At least 7,870 individuals were affected across three states, although the nationwide total remains unknown.

  • At least 7,870 individuals were affected across Massachusetts, South Carolina and Rhode Island, with additional cases reported in California and Illinois.
  • Hackers used compromised Fireworks AI credentials to access Uber data, including drivers’ license information.
  • Notion also reported exposure of AI chat transcripts following the Fireworks incident. Uber is offering affected individuals free credit monitoring and fraud assistance.

Fireworks notified Uber on October 9 that unauthorized parties had obtained credentials and used them to access the company’s account. Uber disconnected its data integrations with the provider and launched an investigation.

By October 13, investigators had confirmed that the attackers accessed data containing drivers’ personal information.

The compromised records included driver’s license information, creating potential identity theft and fraud risks.

Uber confirms at least 7,870 individuals affected

State disclosures identify 3,430 affected individuals in South Carolina, 2,567 in Massachusetts and 1,873 in Rhode Island.

Uber also notified drivers in California and Illinois but has not disclosed how many residents were affected in those states.

US stateAffected individuals
South Carolina3,430
Massachusetts2,567
Rhode Island1,873
CaliforniaNot disclosed
IllinoisNot disclosed
Confirmed total7,870

Additional breach notifications included information for residents of the District of Columbia, Maryland, North Carolina, New York and Pennsylvania. Uber has not identified evidence that the stolen information has been misused.

The company is providing affected individuals with free credit monitoring and fraud assistance through Cyberscout, a TransUnion company. Recipients have 90 days from the date of their notification letter to enroll.

Fireworks AI attack exposed customer data and credentials

Fireworks AI provides infrastructure for companies developing and deploying artificial intelligence models.

The provider identified suspicious activity on October 6 involving exposed environment variables, API keys and authentication tokens associated with certain customer evaluation workloads.

A separate incident on October 23-24 involved unauthorized access to Fireworks development systems through a third-party cloud provider.

Potentially exposed information included customer billing records, account metadata, incomplete model weights and datasets.

Fireworks attributed the intrusion to financially motivated cybercriminals targeting credentials used to access large language models.

Investigators have not publicly established which intrusion exposed Uber’s driver information or where the records were stored.

Fireworks said its zero-data-retention inference services were unaffected, although certain batch-processing products require customer datasets to be retained.

Notion AI chat transcripts also compromised

Notion confirmed that attackers accessed AI chat transcripts associated with its Insights feature during the Fireworks security incident.

The compromised dataset affected fewer than 1% of Notion workspaces. The company said most exposed transcripts could not be linked to identifiable individuals or organizations.

Notion subsequently disconnected its services from Fireworks, disabled Insights and advised affected customers to rotate credentials potentially entered into AI conversations.

AI software companies Cursor and Harvey were also reportedly examining possible exposure, although confirmed data theft involving those companies has not been established.

Stolen AI credentials create additional cybersecurity risks

Fireworks linked the attack to criminals seeking to steal and resell AI service access tokens, an activity commonly known as LLMjacking.

Attackers can exploit compromised API keys and authentication credentials to use AI models without authorization, leaving legitimate customers responsible for the resulting computing charges.

Google Threat Intelligence Group and Mandiant have documented AI credential abuse producing more than $100,000 in unauthorized charges within a single day, with cumulative losses reaching $1 mn at some organizations.

Those incidents have not been directly connected to the Fireworks breach.

Uber’s case demonstrates another potential consequence of compromised AI infrastructure credentials: access to sensitive personal information maintained by external providers.

Unlike API keys or passwords, driver’s license numbers cannot be routinely rotated, leaving affected individuals exposed to potentially persistent identity fraud risks.

Uber faces second reported security incident in 2026

The Fireworks breach follows an August security incident involving Uber Freight, the company’s truck shipping business.

An extortion group claimed to have stolen data from that unit, prompting an investigation. Uber also experienced a major breach in 2016 involving personal information belonging to drivers and passengers.

The latest incident involves third-party infrastructure rather than a confirmed direct compromise of Uber’s internal systems.

For companies integrating external AI services, the case raises questions about which customer records are transferred to vendors, how long those records are retained and whether development environments have adequate access controls.

Affected Uber drivers can consider freezing their credit files with Equifax, Experian and TransUnion, monitoring financial accounts and remaining alert to phishing attempts involving their personal information.

Uber has not disclosed a final nationwide count of affected drivers or additional findings establishing the full extent of the compromised records.