Facebook said today some 90 million of its users may get forcibly logged out of their accounts after the company fixed a rather glaring security vulnerability in its Web site that may have let attackers hijack user profiles.

In a short blog post published this afternoon, Facebook said hackers have been exploiting a vulnerability in Facebook’s site code that impacted a feature called “View As,” which lets users see how their profile appears to other people.

“This allowed them to steal Facebook access tokens which they could then use to take over people’s accounts,” Facebook wrote. “Access tokens are the equivalent of digital keys that keep people logged in to Facebook so they don’t need to re-enter their password every time they use the app.”

Facebook said it was removing the insecure “View As” feature, and resetting the access tokens of 50 million accounts that the company said it knows were affected, as well as the tokens for another 40 million users that may have been impacted over the past year.

The company said it was just beginning its investigation, and that it doesn’t yet know some basic facts about the incident, such as whether these accounts were misused, if any private information was accessed, or who might be responsible for these attacks.

Although Facebook didn’t mention this in their post, one other major unanswered question about this incident is whether the access tokens could have let attackers interactively log in to third-party sites as the user. Tens of thousands of Web sites let users log in using nothing more than their Facebook profile credentials. If users have previously logged in at third-party sites using their Facebook profile, there’s a good chance the attackers could have had access to those third-party sites as well.

I have asked for clarification from Facebook on this point and will update this post when and if I receive a response. However, I would have expected Facebook to mention this as a mitigating factor if authorized logins at third-party sites were not impacted.

Update: 4:46 p.m. ET: A Facebook spokesperson confirmed that while it was technically possible that an attacker could have abused this bug to target third-party apps and sites that use Facebook logins, the company doesn’t have any evidence so far that this has happened.

“We have invalidated data access for third-party apps for the affected individuals,” the spokesperson said, referring to the 90 million accounts that were forcibly logged out today and presented with a notification about the incident at the top of their feed.

Original story:
Facebook says there is no need for users to reset their passwords as a result of this breach, although that is certainly an option.

More importantly, it’s a good idea for all Facebook users to review their login activity. This page should let you view which devices are logged in to your account and approximately where in the world those devices are at the moment. That page also has an option to force a simultaneous logout of all devices connected to your account.

Source

In September 2016, over 16GB of logs from a service indicated to be digimon.co.in were obtained, most likely from an unprotected Mongo DB instance. The service ceased running shortly afterwards and no information remains about the precise nature of it. Based on enquiries made via Twitter, it appears to have been a mail service possibly based on PowerMTA and used for delivering spam. The logs contained information including 7.7M unique email recipients (names and addresses), mail server IP addresses, email subjects and tracking information including mail opens and clicks.

Source

In September 2018, security researcher Bob Diachenko discovered a massive collection of personal details exposed in an unprotected Mongo DB instance. The data appears to have been used in marketing campaigns (possibly for spam purposes) but had little identifying data about it other than a description of “Yahoo_090618_ SaverSpy”. The data set provided to HIBP had almost 2.5M unique email addresses (all of which were from Yahoo!) alongside names, genders and physical addresses.

Source

In September 2016, the real estate investment site Real Estate Mogul had a Mongo DB instance compromised and 5GB of data downloaded by an unauthorised party. The data contained real estate listings including addresses and the names, phone numbers and 308k unique email addresses of the sellers. Real Estate Mogul was advised of the incident in September 2018 and stated that they “found no instance of user account credentials like usernames and passwords nor billing information within this file”.

Source

In September 2016, almost 21GB of data from the French website used for “standardised and decentralized means of exchange for publishing newsgroup articles” NemoWeb was leaked from what appears to have been an unprotected Mongo DB. The data consisted of a large volume of emails sent to the service and included almost 3.5M unique addresses, albeit many of them auto-generated. Multiple attempts were made to contact the operators of NemoWeb but no response was received.

Source

Government Payment Service Inc. — a company used by thousands of U.S. state and local governments to accept online payments for everything from traffic citations and licensing fees to bail payments and court-ordered fines — has leaked more than 14 million customer records dating back at least six years, including names, addresses, phone numbers and the last four digits of the payer’s credit card.

Indianapolis-based GovPayNet, doing business online as GovPayNow.com, serves approximately 2,300 government agencies in 35 states. GovPayNow.com displays an online receipt when citizens use it to settle state and local government fees and fines via the site. Until this past weekend it was possible to view millions of customer records simply by altering digits in the Web address displayed by each receipt.

On Friday, Sept. 14, KrebsOnSecurity alerted GovPayNet that its site was exposing at least 14 million customer receipts dating back to 2012. Two days later, the company said it had addressed “a potential issue.”

“GovPayNet has addressed a potential issue with our online system that allows users to access copies of their receipts, but did not adequately restrict access only to authorized recipients,” the company said in a statement provided to KrebsOnSecurity.

The statement continues:

“The company has no indication that any improperly accessed information was used to harm any customer, and receipts do not contain information that can be used to initiate a financial transaction. Additionally, most information in the receipts is a matter of public record that may be accessed through other means. Nonetheless, out of an abundance of caution and to maximize security for users, GovPayNet has updated this system to ensure that only authorized users will be able to view their individual receipts. We will continue to evaluate security and access to all systems and customer records.”

In January 2018, GovPayNet was acquired by Securus Technologies, a Carrollton, Texas- based company that provides telecommunications services to prisons and helps law enforcement personnel keep tabs on mobile devices used by former inmates.

Although its name may suggest otherwise, Securus does not have a great track record in securing data. In May 2018, the New York Times broke the news that Securus’ service for tracking the cell phones of convicted felons was being abused by law enforcement agencies to track the real-time location of mobile devices used by people who had only been suspected of committing a crime. The story observed that authorities could use the service to track the real-time location of nearly any mobile phone in North America.

Just weeks later, Motherboard reported that hackers had broken into Securus’ systems and stolen the online credentials for multiple law enforcement officials who used the company’s systems to track the location of suspects via their mobile phone number.

A story here on May 22 illustrated how Securus’ site appeared to allow anyone to reset the password of an authorized Securus user simply by guessing the answer to one of three pre-selected “security questions,” including “what is your pet name,” “what is your favorite color,” and “what town were you born in”. Much like GovPayNet, the Securus Web site seemed to have been erected sometime in the aughts and left to age ungracefully for years.

Choose wisely and you, too, could gain the ability to look up anyone’s precise mobile location.

Data exposures like these are some of the most common but easily preventable forms of information leaks online. In this case, it was trivial to enumerate how many records were exposed because each record was sequential.

E-commerce sites can mitigate such leaks by using something other than easily-guessed or sequential record numbers, and/or encrypting unique portions of the URL displayed to customers upon payment.

Although fixing these information disclosure vulnerabilities is quite simple, it’s remarkable how many organizations that should know better don’t invest the resources needed to find and fix them. In August, KrebsOnSecurity disclosed a similar flaw at work across hundreds of small bank Web sites run by Fiserv, a major provider of technology services to financial institutions.

In July, identity theft protection service LifeLock fixed an information disclosure flaw that needlessly exposed the email address of millions of subscribers. And in April 2018, PaneraBread.com remedied a weakness that exposed millions of customer names, email and physical addresses, birthdays and partial credit card numbers.

Got a tip about a security vulnerability similar to those detailed above, or perhaps something more serious? Please drop me a note at krebsonsecurity @ gmail.com.

Source

In September 2018, a collection of almost 42 million email address and plain text password pairs was uploaded to the anonymous file sharing service kayo.moe. The operator of the service contacted HIBP to report the data which, upon further investigation, turned out to be a large credential stuffing list. For more information, read about The 42M Record kayo.moe Credential Stuffing Data.

Source

In approximately 2017, the website for Russian speakers in America known as Russian America suffered a data breach. The incident exposed 183k unique records including names, email addresses, phone numbers and passwords stored in both plain text and as MD5 hashes. Russian America was contacted about the breach but did not respond.

Source

In July 2016, the India-based food delivery service FreshMenu suffered a data breach. The incident exposed the personal data of over 110k customers and included their names, email addresses, phone numbers, home addresses and order histories. When advised of the incident, FreshMenu acknowledged being already aware of the breach but stated they had decided not to notify impacted customers.

Source

In approximately December 2016, the online service for World of Warcraft private servers Warmane suffered a data breach. The incident exposed over 1.1M accounts including usernames, email addresses, dates of birth and salted MD5 password hashes. The data was subsequently extensively circulated online and was later provided to HIBP by whitehat security researcher and data analyst Adam Davies.

Source