Skip to content
NewA calmer, faster ZeroHour

16-Year-Old Researcher Discovers Microsoft Authentication Bug Exposing 17.3 Trillion Records

GBHackersby Divya ()3 min read4 · 3 reads
AI summary · grok-4.7Summaries by Vani AI

A teenage researcher found Microsoft Titan accepted unsigned JWTs, enabling admin SQL against huge internal analytics stores.

Researcher Faav, 16, found that Microsoft’s internal Titan analytics API checked JWT claims but did not cryptographically validate signatures. An unsigned token with an admin UPN was accepted, allowing privileged SQL against publicly reachable routes. Metadata from 17 ClickHouse databases implied about 17.3 trillion stored rows, a figure that includes replicas and derived data rather than confirmed customer records. Limited queries exposed roughly 25,000 account and email records and employee metadata; Microsoft restricted the endpoint on September 9, 2026, paid a $5,000 bounty, and no malicious exploitation is reported.

Key facts
  • Titan accepted unsigned JWTs, including alg:none, without verifying signatures.
  • Forged admin claim let the researcher run privileged SQL on an exposed API.
  • 17.3 trillion is an estimated row count, not exfiltrated customer records.
  • Metadata views included about 25,000 accounts and nearly 18,000 employee emails.
  • Microsoft locked the API four days after the Sept. 5 report and paid $5,000.
Analyze this story with Vani AIImpact, who is affected and what defenders should do, in a new chat

From the original article

3 min read at gbhackers.com

A 16-year-old security researcher known as Faav discovered a significant authentication flaw in Microsoft’s internal Titan analytics service. This vulnerability could have let an attacker impersonate an administrator and run unauthorized SQL queries. The researcher estimated that the vulnerable environment contained 17.

GBHackers

By Divya · gbhackers.com

Read the full article at GBHackers

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.