What was exposed
Replaying the Dow Jones May 2017 exposure
- 1
Configured: a bucket open to any AWS account
Not stated; before May 30, 2017A Dow Jones S3 bucket grants download rights to the Authenticated Users group, any holder of a free AWS account; who set the grant and when are not stated.
- 2
Reachable: subscriber records for any AWS account
Until June 6, 2017; start not statedAnyone who registers a free AWS account can download the bucket by its address: subscriber records with names, addresses, login emails and the last four card digits, and a Risk & Compliance list of 1.6 million people and entities.
- 3
Found and closed: secured a day after the notice
May 30–June 6, 2017An UpGuard researcher finds the bucket on May 30 and begins downloading on June 1; UpGuard says it notified Dow Jones on June 5, and the bucket is secured on June 6; who made the change is not stated.
Subscriber records with names, addresses, login emails and the last four card digits, and a Risk & Compliance list of 1.6 million people and entities, were downloadable by any holder of an AWS account until June 6, 2017; when the exposure began is not stated, Dow Jones sees no reason to believe data was taken, and UpGuard does not know whether anyone else read it.
Where cloud-provider controls fit
Amazon S3 Block Public Access on the account, or set for every account by an organization policy, would have refused the grant to the Authenticated Users group. Object Ownership would have closed the ACL path, and the sources do not say whether an ACL or a bucket policy carried the grant; a service control policy would have kept the ACL write to approved administrators. While the grant stood, a resource control policy that keeps your buckets to your organization's principals, or the IgnorePublicAcls setting of the same block, would have kept outside AWS accounts from reading. Each of these, in the form described, dates from after this exposure. Azure has no grant to every account holder and refuses its nearest opening, anonymous access, with an account setting; Google Cloud refuses the same grant with the public access prevention constraint, and Oracle Cloud refuses a public bucket, which is anonymous access, with a Security Zone. Finding an open bucket and closing it stay the owner's job: the provider's checks report such a bucket rather than refuse it.
Could this have been prevented?
In hindsight, not as a whole. 1 control would have refused a step outright, and 4 controls would have limited one. 2 steps have no cloud-provider control. The counts below are control assessments, not attack stages: several can address the same step. The same replay with the controls in place before the incident shows where each phase would have been refused or bounded.
- 1
- Prevented
- 4
- Contained
- 2
- Not covered
How it could have been prevented
The same replay with the controls in place. Green marks where a step is refused, blue where it is bounded, and red where no control reaches it. Open a row to read the control, why it acts, and its caveat.
Replaying the Dow Jones May 2017 exposure
1 · Configured: a bucket open to any AWS account
1 Prevented2 Contained3 on other clouds2 · Reachable: subscriber records for any AWS account
2 Contained3 on other clouds3 · Found and closed: secured a day after the notice
2 Not coveredKey facts
- An Amazon S3 bucket of Dow Jones & Company granted download rights to the Authenticated Users group, which Amazon defined as any user with an AWS account; registration for such an account was free
- Which object carried the grant, who set it and when were not stated; Dow Jones called the cause an internal error, not a hack or attack, and said the content was over-exposed on Amazon Cloud, not the open internet
- The bucket held four compressed subscriber files totaling 771 MB, with names, addresses, login emails and the last four digits of card numbers, and a Risk & Compliance file of 1.6 million rows of people or entities
- Dow Jones confirmed at least 2.2 million affected customers; UpGuard estimated up to four million, and said duplicate subscriptions may explain part of the gap
- An UpGuard researcher found the bucket on the evening of May 30, 2017 and began downloading on June 1; UpGuard says it notified Dow Jones on June 5, and the bucket was secured on June 6; one news report dated the discovery to May 31
- Dow Jones said it saw no reason to believe any data was taken, and did not say whether logs supported that; UpGuard did not know whether anyone else accessed the bucket
Sources
Every fact in this breakdown comes from the public reporting below. The replay is a model of that reporting, not a reconstruction of the environment it describes.
- 1.Cloud Leak: WSJ Parent Company Dow Jones Exposed Customer DataUpGuard · Jul 17, 2017 · ResearchAuthenticated Users download grant on an Amazon S3 bucket, Amazon's definition of the group, free AWS account, owner Dow Jones & Company confirmed by its CISO, subscriber files and fields, Risk & Compliance list of 1.6 million rows, several dozen directories, discovery on May 30, 2017, download from June 1, 2017, closure on June 6, 2017, Dow Jones count of 2.2 million, UpGuard estimate of up to four million
- 2.Hole in the Cloud Service Bucket: Dow Jones Data ExposedGovInfoSecurity (ISMG) · Jul 19, 2017 · NewsDow Jones statement on cause, reach and access, notification on June 5, 2017 as UpGuard put it, bucket secured the next day, Amazon statement that stored data is by default accessible only to the account holder
- 3.Dow Jones Leaks Personal Info of 2.2 Million CustomersInfosecurity Magazine · Jul 17, 2017 · NewsDow Jones confirmation of at least 2.2 million, acknowledgment in the Wall Street Journal on July 16, 2017
- 4.Yet Another Misconfigured Amazon S3 Bucket Exposes Dow Jones Customer DataAWS Insider (Converge 360) · Jul 19, 2017 · Newsaccess by any holder of a free AWS account described as semi-public, AWS notice to customers about bucket ACLs open to any user
The estate in the replay is illustrative and drawn at the level of the exposure; the owner's real environment is not public. Verdicts describe what a control does when it is in place before the data is exposed. They are not an assessment of the owner's security program.