Skip to main content

full timeline

Fireworks AI supply chain breach

A security failure at third-party infrastructure provider Fireworks AI has exposed the personal data of thousands of Uber drivers.

Factlen Newsroom· Updated

Follow Fireworks AI supply chain breach in the Factlen app and get an alert when there is news. Get the app

Illustration: Uber confirmed that thousands of driver's licenses were exposed in a third-party data breach.From: Hackers Steal Thousands of Uber Driver Licenses Through Fireworks AI Supply Chain Breach

Share this story

Latest report

Hackers Steal Thousands of Uber Driver Licenses Through Fireworks AI Supply Chain Breach

A security failure at third-party infrastructure provider Fireworks AI has exposed the personal data of thousands of Uber drivers. The breach highlights the growing vulnerability of enterprise software supply chains as companies rush to integrate generative AI tools.

  • Hackers stole thousands of Uber driver's licenses by breaching third-party AI infrastructure provider Fireworks AI.
  • The exposure stems from a stolen API key that previously led to a data breach at the AI coding platform Cursor.
  • Uber filed a formal notice of the security incident with the California Department of Justice on November 10, 2026.
  • The incident highlights severe supply chain vulnerabilities as enterprises rush to integrate generative AI tools.

Uber drivers across the United States are facing a new wave of identity theft risks following a confirmed data breach. The ride-hailing company acknowledged that hackers stole thousands of driver's license records. However, the intrusion did not occur on Uber's own internal servers.[1][2]

Instead, the exposure stems from a compromised third-party vendor known as Fireworks AI. The startup, which provides generative artificial intelligence infrastructure, suffered a security failure that cascaded up the supply chain. Uber filed a formal notice of the breach with the California Department of Justice on November 10, 2026.[4][5]

The incident highlights a growing vulnerability in the technology sector as companies rush to integrate AI tools. While startups often market their platforms as enterprise-ready and highly secure, the underlying architecture frequently relies on simple text-based API keys. When those keys leak, the protective layers vanish entirely.[3][4]

Uber's regulatory filing with the state describes the event plainly as a "Security Incident," though the company has not yet detailed the exact number of affected contractors. Because neither Uber nor Fireworks AI has provided direct quotations to the press regarding the technical specifics, independent researchers are left to piece together the attack path.[2][5]

Cascading failures in the AI ecosystem

The root cause of the Uber data exposure appears to trace back to a broader security collapse involving Fireworks AI. On November 1, cybersecurity researchers documented that a stolen Fireworks AI key had already led to other significant breaches. That initial leak compromised data at Cursor, a popular AI-assisted coding platform.[6]

Illustration: The breach originated at Fireworks AI, a startup providing generative artificial intelligence infrastructure.

The attackers did not stop at simply scraping data from the compromised servers. They utilized the stolen credentials to distribute a trojanized version of a Claude code extension. This malicious software update allowed the hackers to establish persistent access across multiple developer environments.[6]

Fireworks AI positions itself as a high-performance engine for running large language models, promising speed and reliability for corporate clients. Yet, the mechanics of how it handles authentication proved to be a critical weak point. A single compromised access token granted unauthorized users the ability to query sensitive databases.[3][4]

Security analysts note that the integration of third-party AI services often requires companies to share vast amounts of proprietary data to train or fine-tune models. In Uber's case, the data pipeline connected to Fireworks AI apparently included highly sensitive personal identification records. The exact operational reason this specific dataset was accessible via the AI startup remains unclear.[3][4]

The cascading nature of the breach underscores the fragility of modern software supply chains. When a foundational infrastructure provider experiences a leak, every client utilizing their application programming interface becomes a potential target. The attackers simply followed the access privileges upstream to reach Uber's records.[4][6]

Regulatory scrutiny and delayed disclosures

The timeline of the disclosure raises questions about how quickly affected drivers were notified. The initial indicators of the Fireworks AI credential theft surfaced publicly at the beginning of November. However, Uber's formal notification to the California Department of Justice was not filed until ten days later.[5][6]

Uber filed a formal notice of the security incident with the California Department of Justice on November 10.

It took another three days for the breach to reach widespread public awareness, with technology outlets confirming the details on November 13. This delay between the initial third-party compromise and the corporate acknowledgment is a common pattern in supply chain attacks. Companies often require weeks to audit their own logs.[1][4]

The California Department of Justice maintains strict requirements for data breach notifications, particularly when state residents are impacted. Uber's compliance with these regulations forces the incident into the public record, preventing the exposure from being handled quietly behind closed doors. The filing mandates that affected individuals receive formal communication.[5]

The exposure of driver's licenses presents a severe and immediate risk to the affected gig workers. Unlike a compromised credit card, which can be canceled and reissued within days, a driver's license contains static personal information. It includes a home address, date of birth, and physical characteristics.[1][3]

Identity thieves frequently use stolen driver's licenses to open fraudulent bank accounts, apply for loans, or bypass identity verification checks on other platforms. The burden of monitoring for this fraud falls entirely on the victims. Drivers must now actively watch their credit reports for unauthorized activity.[1][2]

The reality of enterprise AI integration

The technology industry is currently experiencing a massive push to embed generative AI into every conceivable workflow. Startups like Fireworks AI are capitalizing on this demand by offering off-the-shelf infrastructure. However, this rapid adoption often outpaces the implementation of robust security protocols.[4]

Marketing materials for AI infrastructure frequently emphasize terms like "enterprise-grade" and "secure by design." The reality, as demonstrated by the Cursor and Uber incidents, is that these systems are only as secure as their weakest authentication token. An exposed key on a developer's machine can unravel millions of dollars in corporate security investments.[6]

Illustration: The initial compromise involved a stolen Fireworks AI key that was later used to distribute a trojanized code extension.

The integration of AI tools introduces new attack vectors that traditional security operations centers are still learning to monitor. When an AI model requires access to a database to generate insights, it creates a persistent bridge between the vendor and the client. If the vendor is compromised, that bridge becomes a highway for data exfiltration.[3][6]

In the wake of the breach, cybersecurity experts are urging corporations to reevaluate their data sharing agreements with AI startups. The principle of least privilege—where a service is only granted the minimum access necessary to function—is often ignored during rapid deployment phases. Uber's exposure suggests that Fireworks AI had access to data far beyond simple test environments.[1][4]

The incident also highlights the specific dangers of the "API economy," where disparate software systems constantly communicate in the background. A single compromised node can silently siphon data for weeks before anomalous traffic is detected. The attackers behind the Fireworks AI breach demonstrated a sophisticated understanding of these interconnected networks.[4][6]

Consequences for the gig economy

For the thousands of Uber drivers whose information was stolen, the breach represents a tangible threat to their financial security. Gig economy workers are particularly vulnerable to identity theft, as their livelihoods depend on maintaining clear background checks and valid documentation.[1][3]

A fraudulent flag on a driver's record can result in immediate suspension from ride-hailing platforms. Resolving these issues often takes months of navigating automated customer service systems. During that time, the driver is entirely cut off from their primary source of income.[3]

Uber has historically faced significant scrutiny over its data security practices, having suffered multiple high-profile breaches over the past decade. While this latest incident originated with a third-party vendor, the ultimate responsibility for protecting driver data rests with the platform that collected it. Regulators are increasingly holding primary corporations accountable for their vendors' failures.[1][5]

Illustration: Gig economy workers face severe identity theft risks when static documents like driver's licenses are exposed.

The fallout from the Fireworks AI compromise is likely to extend beyond Uber and Cursor. As security researchers continue to analyze the trojanized Claude extension and the stolen API keys, additional victims may be identified. The full scope of the data exfiltration remains an active area of investigation.[4][6]

The broader implications for the artificial intelligence industry are significant. Investors have poured billions of dollars into infrastructure providers like Fireworks AI, betting on their ability to scale rapidly. However, if these platforms cannot guarantee the security of client data, enterprise adoption may face severe headwinds.[3]

Moving forward, the technology sector must reconcile the demand for rapid AI deployment with the fundamental requirements of data protection. Until third-party vendors can demonstrate verifiable security architectures that isolate client data, supply chain attacks will remain a primary threat.[4]

For the affected Uber drivers, that realization offers little comfort as they monitor their credit reports for signs of fraud. The next phase of this incident will depend on how aggressively state regulators pursue enforcement actions against the entities involved.[1][5]

Sources

  1. [1]TechCrunch — Uber confirms hackers stole thousands of drivers' license information in data breach Read on TechCrunch →
  2. [2]Reuters — Uber says US drivers' license data stolen in data breach Read on Reuters →
  3. [3]Gizmodo — Hackers Stole Thousands of Uber Drivers' Data Through an AI Startup Read on Gizmodo →
  4. [4]PCMag — Uber Says Hackers Stole Driver's License Data Via Third-Party Fireworks AI Read on PCMag →
  5. [5]State of California Department of Justice — Uber Technologies Notice of Security Incident Read on State of California Department of Justice →
  6. [6]The Hacker News — Stolen Fireworks AI Key Led to Cursor Data Breach and Trojanized Claude Code Extension Read on The Hacker News →

Charts, maps and images

The incident highlights the risks of relying on third-party AI infrastructure providers.From: Fireworks AI Security Breach Exposes Notion AI Customer Chat Transcripts ·

How the story developed

The security vulnerabilities surrounding the third-party AI infrastructure provider Fireworks AI first involved the AI coding platform Cursor, which experienced a data breach linked to a stolen API key.

On November 10, 2026, Uber filed a formal notice of a security incident with the California Department of Justice regarding the exposure of driver data.

How Notion AI prompts were exposed in the Fireworks AI breach.From: Fireworks AI Security Breach Exposes Notion AI Customer Chat Transcripts ·

By November 12, 2026, a cyberattack on Fireworks AI exposed the chat transcripts of Notion AI users. This data exposure occurred despite Notion's stated zero-data-retention policy, which was intended to ensure that user prompts were immediately deleted.

Following the exposure of its customer chat logs, Notion paused all traffic to Fireworks AI. The company began routing its requests to alternative AI infrastructure providers to mitigate the security risks associated with third-party model hosting.

Fireworks AI hosts and serves open-source models for enterprise clients.From: Fireworks AI Security Breach Exposes Notion AI Customer Chat Transcripts ·

By November 15, 2026, the breach involving Uber was established to have exposed thousands of driver's licenses. Hackers accessed the personal data through the Fireworks AI supply chain using the same stolen API key responsible for the earlier Cursor breach.