149,000 Uber drivers exposed through a third-party AI provider: how a single set of credentials led to a major identity risk

Uber is notifying 149,346 drivers that an unauthorised party accessed their personal information through Fireworks AI, a cloud platform that runs and fine-tunes open-source AI models. The attacker used compromised credentials belonging to the provider to reach data Uber had shared with it. The exposed data includes names, addresses, dates of birth, driver's licence numbers, other government ID numbers and images of driver's licences.

AI Security News 5 min read

149,000 Uber drivers exposed through a third-party AI provider: how a single set of credentials led to a major identity risk

Uber shares data with third-party service providers so that they can perform work on its behalf. In this case, the provider was Fireworks AI. Uber did not name the provider in its breach notifications, but a company spokesperson confirmed the link to Fireworks AI.

The timeline, as Uber reported it, runs as follows:

Date

Event

7 October 2026

Attackers use compromised service provider credentials to access data Uber had shared with the provider

9 October 2026

Uber learns of the incident and cuts its data connections with the provider

13 October 2026

Uber's investigation determines what data was accessed

9 to 10 November 2026

Uber updates its disclosure and files breach notifications with US state authorities

Uber has said the compromised credentials did not give access to its other systems. The company has filed breach notifications in California, Iowa, Massachusetts, South Carolina, Texas and Vermont, and has offered affected drivers free credit monitoring and fraud assistance.

What was taken

According to Uber's notifications, the stolen information includes:

  • names and addresses

  • dates of birth

  • driver's licence numbers

  • other government-issued identification numbers

  • images of driver's licences

In a November update, Uber said the breach also affected account and billing information, some customer datasets and incomplete AI models. Reports on the full scope of these additional categories differ, and Uber has not published a complete list, so this article treats them as reported rather than confirmed.

Uber has said it has not identified any misuse of the data. Notifying drivers before any misuse is found is a deliberate choice, because stolen identity documents are often used long after a breach.

Why this matters for drivers

Driver's licence numbers and dates of birth are valuable to criminals because they are hard to change. A password can be reset, but a licence number remains the same. Combined with a name, address and date of birth, this information can support identity fraud, such as opening accounts or applying for credit in someone else's name.

Images of licences are a particular concern, because they can be used to create fake documents that pass basic checks. Drivers should expect targeted phishing emails and phone calls that refer to Uber, the breach or their licence details.

Why it matters for organisations

The attacker did not break into Uber's own systems, according to Uber. They used credentials belonging to a provider, which then gave access to data Uber had shared. This pattern is common in supply-chain incidents: a company can have strong defences but still lose data held by its suppliers.

Uber's own filings show that it has been aware of this risk for some time. Its annual reports describe third-party providers as a source of security exposure, and the company has previously faced a large breach. A 2016 incident exposed data from about 57 million drivers and consumers, including the licence numbers of about 600,000 drivers, and led to a regulatory order from the US Federal Trade Commission in 2023.

The lesson for other organisations is practical: sharing data with a provider is a security decision, not only a contractual one. Each provider holding sensitive personal data should be treated as an extension of your own attack surface.

What drivers should do

  1. Watch for unexpected messages that mention Uber, a breach, a licence or a government agency. Do not click links or give information over the phone unless you contacted the organisation yourself.

  2. Enrol in the credit monitoring that Uber offers, using the link or number in the official notice.

  3. Check your credit report regularly, and consider a fraud alert or credit freeze, which are available in many jurisdictions at no cost.

  4. Consider replacing your licence if you are worried about the number's misuse, and ask your licensing authority what options exist.

  5. Use strong, unique passwords and enable multi-factor authentication on accounts linked to your email address.

What organisations should do

Companies that share personal data with suppliers should verify that each supplier restricts credentials to the minimum access needed, enforces multi-factor authentication, and monitors for unusual access. Contracts should set clear expectations for breach notification, and organisations should be able to cut off a provider's data access quickly, as Uber did.

Data minimisation matters too. Sharing only the fields a provider needs reduces the damage when a supplier is compromised. Storing identity document images with a third party is a higher-risk choice that should be justified.

The lesson

The Uber case shows that a single compromised set of supplier credentials can expose sensitive identity data for many people at once. Uber acted quickly once it learned of the breach, but the data was already gone, and the drivers affected will carry the risk for years.

The question for any organisation is not only whether its own systems are secure, but whether every provider holding its data would contain a breach as quickly as Uber did.

Blog

Microsoft 4 min read

Fake "Outage" emails follow Microsoft 365 disruption: check for held-message lures before you click

On 12 November 2026, Microsoft 365 and Outlook suffered a widespread outage. Downdetector recorded user reports from 74 countries and territories. Microsoft says the disruption was not caused by a cyberattack. Within a day, attackers began sending phishing emails that falsely claim messages were held during the outage and push recipients to a "restore messages" link. Microsoft has warned that these emails resemble credential phishing, but has not confirmed that any credentials were stolen.

Microsoft 4 min read

Outlook under attack: Microsoft's November patch closes an exploited flaw that leaks NTLMv2 hashes

On 10 November 2026, Microsoft released fixes for more than a thousand vulnerabilities. One of them, CVE-2026-124194 in Outlook, was already being exploited by attackers before Microsoft had a patch ready. Opening a crafted message file can expose a user's password hash, which an attacker can then use to impersonate that user. Every device running Outlook needs the update.

Security 4 min read

Autopay confirms unauthorised access to payment infrastructure: what is known and what customers should check

Autopay, a Polish online payment operator, has confirmed that an unauthorised party gained access to part of its infrastructure and to some payment-related data. The company detected the intrusion on 11 November 2026 and announced it on 13 November. It has not said whether data was copied, how many customers are affected, or which data categories were involved. Customers should check their bank statements now, because the company itself cannot rule out data loss.