"Life is short. Have an affair®." This is the (in)famous marketing slogan used by Ashley Madison, a Canadian web site founded in 2008 and operated by Avid Life Media Inc. with the explicit mission statement of helping married individuals chat, connect and ultimately have affairs with one another. The site assured users that use of its services would be "anonymous" and "100 per cent discreet," but, unfortunately, this was not to be the case.
Between July 15 and Aug. 20, 2015, a person/group identifying itself as "The Impact Team" hacked ALM and published details, initially on the Darkweb and eventually on the open web, of approximately 36 million user accounts. Leaked data included profile information (user names, addresses, passwords, phone numbers, the types of experiences they were looking for on the site, gender, height, weight, ethnicity, body type); account information used to facilitate access to the Ashley Madison service (e-mail addresses, security questions, hashed passwords); and billing information (billing addresses and the last four digits of credit card numbers); in addition to ALM internal documents and the CEO's private e-mail messages. User information was quickly disseminated through several public web sites. Despite the best efforts of ALM's counsel to quickly shut down the spread of data using DMCA copyright notices after the material appeared on Twitter and other social media sites, the breached information continued to be publicly searchable.
The fallout was swift. Reports of suicides in Canada and the U.S., myriad job resignations and marital breakups surfaced, arising from the data exposure and related public shaming. In Alabama, editors at one newspaper decided to print all the names of people from the region who appeared on the Ashley Madison database. Scammers and extortionists have also targeted Ashley Madison's users (and alleged users) on a global basis, falsely claiming they could remove a user's information from published data or threatening to publicly shame users online unless they sent a ransom payoff in Bitcoins to the blackmailers. Malware may have also been delivered through web sites offering to scrub user information from stolen data lists.
Further, on Aug. 20, 2015, Charney Lawyers and Sutts Strosberg LLP launched a national class action against Avid Dating Life and ALM in Toronto on behalf of all residents of Canada who subscribed to the web site. The plaintiff claimed $760 million in damages. On Aug. 25, 2015, the same law firms filed a class action in Montreal on behalf of all Quebec residents who subscribed to Ashley Madison for failure to protect their privacy and the dubious "paid delete" process that will be further discussed below.
Given the scale of the data breach, the sensitivity of the information involved, the impact on affected individuals and the international reach of ALM's business, the Office of the Privacy Commissioner of Canada and the Office of the Australian Information Commissioner jointly investigated ALM's privacy practices at the time of the data breach and on Aug. 22, 2016, published a detailed and damning report that found ALM had contravened the Canadian Personal Information Protection and Electronic Documents Act in myriad ways.
While space does not permit a detailed analysis of all of ALM's contraventions of PIPEDA, I will focus on five key takeaways/"lessons learned" that will be useful for all operators of web sites, not just those targeted at adulterers.
(i) Get serious about your security. For a site whose business was awash in personal information, ALM had remarkably mediocre security practices. The OPC found definite gaps in its critical security coverage, including a failure to implement commonly used detective countermeasures, such as intrusion detection systems or prevention systems that could facilitate detection of attacks or identify anomalies (i.e., ALM failed to monitor unusual logins). There were instances of unauthorized access to ALM's systems, using valid security credentials, weeks before the actual data disclosure, which gave further credence to the finding that ALM was not adequately monitoring its systems for intrusion. ALM had no documented risk management framework guiding how it could determine what security measures would be appropriate for the privacy risks it faced to ensure its security arrangements were adequate for its business purposes, resulting in holes. As the OPC's own press release noted:
Not surprisingly, the OPC found ALM's security safeguards were insufficient or absent at the time of the data breach. As the OPC noted, any organization that holds large amounts of PI must have safeguards appropriate to the sensitivity and amount of information collected, supported by an adequate information security governance framework that is often reviewed and updated, to ensure practices appropriate to the risks are consistently understood and effectively implemented. The lack of such framework was unacceptable and failed to prevent "multiple security weaknesses."
Interestingly, ALM argued it could not have the same level of documented compliance frameworks as larger and more sophisticated organizations. However, the OPC dismissed this argument, stating that ALM should have implemented a comprehensive security program given: (i) the quantity and nature of personal information that it held; (ii) the foreseeable adverse impact on individuals should their personal information be compromised; and (iii) the representatives that ALM made to its users about security and discretion. So being a smaller company does not provide any excuse for bad security practices and businesses must take the time and spend the necessary monies to invest in security appropriately.
(ii) Document, document, document. At the time of the data breach, ALM did not have documented information security policies or practices for managing network permissions -- its director of information security had only been engaged since early 2015 and was in the process of developing written security procedures and document when the hack occurred. This clearly worked against Ashley Madison since ALM's employees were implementing undocumented security policies. ALM had also just started training its staff on general privacy and security a few months before the breach and approximately 75 per cent of personnel had not been trained at the time of the incident.
The takeaway here is clear: Organizations that hold personal information electronically must adopt clear and appropriate processes, procedures and systems to handle information security risks, supported by internal or external expertise. Organizations that deal in sensitive personal information should have, at a minimum: (i) security policy(ies); (ii) explicit risk management process that addresses information security matters, drawing on adequate expertise; and (iii) adequate privacy and security training for all staff. As the OPC noted in its findings, the documentation of privacy and security practices can itself be part of establishing security safeguards.
(iii) Don't lie about your credentials. The OPC found that Ashley Madison was well aware of the sensitivity of the personal information it held and, accordingly, actively marketed to customers that its site was both secure and discreet. At the time of the breach, the front page of the web site included a series of fictitious "trustmarks," which suggested a high level of security and discretion, including a medal icon labelled "trusted security award," a lock icon indicating the web site was "SSL secure" and a statement that the web site offered a "100 per cent discreet" service. These statements were found to convey a general impression that the site held a high standard of security and that individuals could rely on these assurances.
The OPC found that while ALM did provide some information about its security safeguards and account closure options, critical elements of its practices that would have been material to prospective users' decision to join Ashley Madison were either absent, difficult to understand or deceptive, including the fabricated trustmarks rather than validated designations of third parties. The OPC found that some individuals may not have chosen to share their personal information with ALM had they not been misled at registration by the fake security trustmarks and if they had been aware their information would be retained indefinitely unless they paid a fee for deletion and the posting of such fabricated marks was found to invalidate user consent upon user signup. The message here: False or misleading statements may impact the validity of a user's consent to the sharing/disclosure of their personal information.
(iv) Mean what you say. At the time of the hack, ALM's terms of use warned users that security or privacy information could not be guaranteed, and if they accessed or transmitted any content through the use of the Ashley Madison service, they did so at their sole discretion and at their sole risk. The terms also contained significant disclaimers and a release in favour of ALM, its various corporate entities, shareholders, officers, directors, employees, etc., relating to disclosure or use of personal information by third parties. However, the OPC was not impressed by the disclaimer and found that: (i) this statement could not absolve ALM of its legal obligations under PIPEDA; and (ii) considering the nature of the personal information collected by ALM, its level of security safeguards should have been high in accordance with PIPEDA's requirements.
Even after a careful reading of the both the ALM privacy policy and terms of use, it was unclear to me that, unless a user chooses a full deletion, their profile would be retained by ALM indefinitely. Moreover, users were not informed until after they had paid for the full delete that their information would be retained for an additional 12 months. Accordingly, ALM was ordered by the OPC to review (and improve) its terms and conditions, privacy policy and other information for accuracy and clarity regarding its information-handling practices, including making it clear how to deactivate accounts and providing the details of all the deactivation and deletion options available, and to ensure it does not make misleading representations. The takeaway here from the OPC: Omissions or lack of clarity of material statements (including in privacy policies and terms of use) can also impact the validity of users' consent to share and disclose their personal information.
(v) Have reasonable retention practices. PIPEDA places limits on the length of time personal information may be retained -- it should only be retained as long as necessary to fulfill the purpose for which it was collected. Organizations are required to develop guidelines that include minimum and maximum retention periods for personal information, and personal information that is no longer required must be destroyed, erased or made anonymous.
At the time of the breach, Ashley Madison offered its users two ways to close a user account -- the first was a free "basic deactivation" (which removed the user's profile from search results but allowed profile information and messages to be accessible to members with which users had communicated). Following account deactivation, information associated with the account was retained by Ashley Madison indefinitely. ALM claimed this was necessary because many users chose to reactivate their accounts after termination (although evidence showed 99.9 per cent of users that chose to reactivate their accounts did so within 29 days of deactivating their account). Information associated with inactive accounts was also retained indefinitely.
Ashley Madison charged $19 for the paid "full delete" service, which promised the deletion of all messages sent or received on the site, browsing history and user data (i.e., photos, site usage history, "personally identifiable information") within 48 hours. According to Ars Technica, at the time of the hack, between 8,000 and 18,000 people had been requesting this feature each month, netting the company more than $1 million from the "full delete" feature alone. Significantly, neither the Ashley Madison privacy policy or terms and conditions contained any notice that a fee would be charged by ALM to delete users' personal information (this policy has since been changed). Despite its written promises, at the time of the breach, ALM kept a user's data for a further 12 months even after they paid for the "full delete" and users were only told their data would be kept "due to legal and financial reasons" (such as processing chargebacks) after ALM accepted the user's payment.
The OPC found that while ALM is entitled to retain information following a basic deactivation for a reasonable period of time to allow for the return of users to its web site, ALM's practice of indefinite retention, and its failure to establish maximum retention periods associated with deactivated user accounts, contravened PIPEDA. Similar considerations apply for inactive accounts. As the OPC noted, retention policies should be based on a "demonstrable rationale and timeline."
The OPC was also deeply unhappy that ALM was charging individuals fees to have their account profile permanently deleted from ALM's databases. PIPEDA is silent on whether a fee can be charged for the withdrawal of consent, but since the fee was not communicated nor available to prospective and existing users in the messaging and contractual terms and conditions between ALM and individuals at the critical point of signup, this practice was found to be a contravention of the PIPEDA principle relating to the withdrawal of consent. The OPC has plainly said there is a high bar associated with charging a fee for deletion and, in order for such fees to be reasonable, further evaluation of factors such as the actual cost to the organization relative to the fee charged or the likely influence it would have on the individual's decision on whether to withdraw consent is required and, regardless, companies must communicate these fees prior to an individual providing consent.
As a result of this investigation, ALM (which has now changed its name to Ruby Corp.) and the OPC entered into a compliance agreement whereby ALM ultimately agreed to clean up its act in a number of areas, including taking stringent remedial measures regarding safeguards, retention, accuracy and transparency.
ALM clearly has a lot of homework to do as many of these actions are required to be fulfilled by the end of the year with others to be completed by May 31, 2017. It remains to be seen whether ALM will meet all of the requirements set out in the compliance agreement, but, in the meantime, prospective users of Ashley Madison may want to think twice about using ALM's various web sites. Buyer beware.
Between July 15 and Aug. 20, 2015, a person/group identifying itself as "The Impact Team" hacked ALM and published details, initially on the Darkweb and eventually on the open web, of approximately 36 million user accounts. Leaked data included profile information (user names, addresses, passwords, phone numbers, the types of experiences they were looking for on the site, gender, height, weight, ethnicity, body type); account information used to facilitate access to the Ashley Madison service (e-mail addresses, security questions, hashed passwords); and billing information (billing addresses and the last four digits of credit card numbers); in addition to ALM internal documents and the CEO's private e-mail messages. User information was quickly disseminated through several public web sites. Despite the best efforts of ALM's counsel to quickly shut down the spread of data using DMCA copyright notices after the material appeared on Twitter and other social media sites, the breached information continued to be publicly searchable.
The fallout was swift. Reports of suicides in Canada and the U.S., myriad job resignations and marital breakups surfaced, arising from the data exposure and related public shaming. In Alabama, editors at one newspaper decided to print all the names of people from the region who appeared on the Ashley Madison database. Scammers and extortionists have also targeted Ashley Madison's users (and alleged users) on a global basis, falsely claiming they could remove a user's information from published data or threatening to publicly shame users online unless they sent a ransom payoff in Bitcoins to the blackmailers. Malware may have also been delivered through web sites offering to scrub user information from stolen data lists.
Further, on Aug. 20, 2015, Charney Lawyers and Sutts Strosberg LLP launched a national class action against Avid Dating Life and ALM in Toronto on behalf of all residents of Canada who subscribed to the web site. The plaintiff claimed $760 million in damages. On Aug. 25, 2015, the same law firms filed a class action in Montreal on behalf of all Quebec residents who subscribed to Ashley Madison for failure to protect their privacy and the dubious "paid delete" process that will be further discussed below.
Given the scale of the data breach, the sensitivity of the information involved, the impact on affected individuals and the international reach of ALM's business, the Office of the Privacy Commissioner of Canada and the Office of the Australian Information Commissioner jointly investigated ALM's privacy practices at the time of the data breach and on Aug. 22, 2016, published a detailed and damning report that found ALM had contravened the Canadian Personal Information Protection and Electronic Documents Act in myriad ways.
While space does not permit a detailed analysis of all of ALM's contraventions of PIPEDA, I will focus on five key takeaways/"lessons learned" that will be useful for all operators of web sites, not just those targeted at adulterers.
(i) Get serious about your security. For a site whose business was awash in personal information, ALM had remarkably mediocre security practices. The OPC found definite gaps in its critical security coverage, including a failure to implement commonly used detective countermeasures, such as intrusion detection systems or prevention systems that could facilitate detection of attacks or identify anomalies (i.e., ALM failed to monitor unusual logins). There were instances of unauthorized access to ALM's systems, using valid security credentials, weeks before the actual data disclosure, which gave further credence to the finding that ALM was not adequately monitoring its systems for intrusion. ALM had no documented risk management framework guiding how it could determine what security measures would be appropriate for the privacy risks it faced to ensure its security arrangements were adequate for its business purposes, resulting in holes. As the OPC's own press release noted:
- There were inadequate authentication processes for employees accessing the company's system remotely as ALM failed to use multi-factor authentication practices.
- ALM's network protections included encryption on all web communications between the company and its users; however, encryption keys were stored as plain, clearly identifiable text on ALM systems. That left information encrypted using those keys at risk of unauthorized disclosure.
- ALM had poor key and password management practices. For example, the company's "shared secret" for its remote access server was available on the ALM Google drive -- meaning anyone with access to any ALM employee's drive on any computer, anywhere, could have potentially discovered it.
- Instances of storage of passwords as plain, clearly identifiable text in e-mails and text files were also found on the company's systems.
Not surprisingly, the OPC found ALM's security safeguards were insufficient or absent at the time of the data breach. As the OPC noted, any organization that holds large amounts of PI must have safeguards appropriate to the sensitivity and amount of information collected, supported by an adequate information security governance framework that is often reviewed and updated, to ensure practices appropriate to the risks are consistently understood and effectively implemented. The lack of such framework was unacceptable and failed to prevent "multiple security weaknesses."
Interestingly, ALM argued it could not have the same level of documented compliance frameworks as larger and more sophisticated organizations. However, the OPC dismissed this argument, stating that ALM should have implemented a comprehensive security program given: (i) the quantity and nature of personal information that it held; (ii) the foreseeable adverse impact on individuals should their personal information be compromised; and (iii) the representatives that ALM made to its users about security and discretion. So being a smaller company does not provide any excuse for bad security practices and businesses must take the time and spend the necessary monies to invest in security appropriately.
(ii) Document, document, document. At the time of the data breach, ALM did not have documented information security policies or practices for managing network permissions -- its director of information security had only been engaged since early 2015 and was in the process of developing written security procedures and document when the hack occurred. This clearly worked against Ashley Madison since ALM's employees were implementing undocumented security policies. ALM had also just started training its staff on general privacy and security a few months before the breach and approximately 75 per cent of personnel had not been trained at the time of the incident.
The takeaway here is clear: Organizations that hold personal information electronically must adopt clear and appropriate processes, procedures and systems to handle information security risks, supported by internal or external expertise. Organizations that deal in sensitive personal information should have, at a minimum: (i) security policy(ies); (ii) explicit risk management process that addresses information security matters, drawing on adequate expertise; and (iii) adequate privacy and security training for all staff. As the OPC noted in its findings, the documentation of privacy and security practices can itself be part of establishing security safeguards.
(iii) Don't lie about your credentials. The OPC found that Ashley Madison was well aware of the sensitivity of the personal information it held and, accordingly, actively marketed to customers that its site was both secure and discreet. At the time of the breach, the front page of the web site included a series of fictitious "trustmarks," which suggested a high level of security and discretion, including a medal icon labelled "trusted security award," a lock icon indicating the web site was "SSL secure" and a statement that the web site offered a "100 per cent discreet" service. These statements were found to convey a general impression that the site held a high standard of security and that individuals could rely on these assurances.
The OPC found that while ALM did provide some information about its security safeguards and account closure options, critical elements of its practices that would have been material to prospective users' decision to join Ashley Madison were either absent, difficult to understand or deceptive, including the fabricated trustmarks rather than validated designations of third parties. The OPC found that some individuals may not have chosen to share their personal information with ALM had they not been misled at registration by the fake security trustmarks and if they had been aware their information would be retained indefinitely unless they paid a fee for deletion and the posting of such fabricated marks was found to invalidate user consent upon user signup. The message here: False or misleading statements may impact the validity of a user's consent to the sharing/disclosure of their personal information.
(iv) Mean what you say. At the time of the hack, ALM's terms of use warned users that security or privacy information could not be guaranteed, and if they accessed or transmitted any content through the use of the Ashley Madison service, they did so at their sole discretion and at their sole risk. The terms also contained significant disclaimers and a release in favour of ALM, its various corporate entities, shareholders, officers, directors, employees, etc., relating to disclosure or use of personal information by third parties. However, the OPC was not impressed by the disclaimer and found that: (i) this statement could not absolve ALM of its legal obligations under PIPEDA; and (ii) considering the nature of the personal information collected by ALM, its level of security safeguards should have been high in accordance with PIPEDA's requirements.
Even after a careful reading of the both the ALM privacy policy and terms of use, it was unclear to me that, unless a user chooses a full deletion, their profile would be retained by ALM indefinitely. Moreover, users were not informed until after they had paid for the full delete that their information would be retained for an additional 12 months. Accordingly, ALM was ordered by the OPC to review (and improve) its terms and conditions, privacy policy and other information for accuracy and clarity regarding its information-handling practices, including making it clear how to deactivate accounts and providing the details of all the deactivation and deletion options available, and to ensure it does not make misleading representations. The takeaway here from the OPC: Omissions or lack of clarity of material statements (including in privacy policies and terms of use) can also impact the validity of users' consent to share and disclose their personal information.
(v) Have reasonable retention practices. PIPEDA places limits on the length of time personal information may be retained -- it should only be retained as long as necessary to fulfill the purpose for which it was collected. Organizations are required to develop guidelines that include minimum and maximum retention periods for personal information, and personal information that is no longer required must be destroyed, erased or made anonymous.
At the time of the breach, Ashley Madison offered its users two ways to close a user account -- the first was a free "basic deactivation" (which removed the user's profile from search results but allowed profile information and messages to be accessible to members with which users had communicated). Following account deactivation, information associated with the account was retained by Ashley Madison indefinitely. ALM claimed this was necessary because many users chose to reactivate their accounts after termination (although evidence showed 99.9 per cent of users that chose to reactivate their accounts did so within 29 days of deactivating their account). Information associated with inactive accounts was also retained indefinitely.
Ashley Madison charged $19 for the paid "full delete" service, which promised the deletion of all messages sent or received on the site, browsing history and user data (i.e., photos, site usage history, "personally identifiable information") within 48 hours. According to Ars Technica, at the time of the hack, between 8,000 and 18,000 people had been requesting this feature each month, netting the company more than $1 million from the "full delete" feature alone. Significantly, neither the Ashley Madison privacy policy or terms and conditions contained any notice that a fee would be charged by ALM to delete users' personal information (this policy has since been changed). Despite its written promises, at the time of the breach, ALM kept a user's data for a further 12 months even after they paid for the "full delete" and users were only told their data would be kept "due to legal and financial reasons" (such as processing chargebacks) after ALM accepted the user's payment.
The OPC found that while ALM is entitled to retain information following a basic deactivation for a reasonable period of time to allow for the return of users to its web site, ALM's practice of indefinite retention, and its failure to establish maximum retention periods associated with deactivated user accounts, contravened PIPEDA. Similar considerations apply for inactive accounts. As the OPC noted, retention policies should be based on a "demonstrable rationale and timeline."
The OPC was also deeply unhappy that ALM was charging individuals fees to have their account profile permanently deleted from ALM's databases. PIPEDA is silent on whether a fee can be charged for the withdrawal of consent, but since the fee was not communicated nor available to prospective and existing users in the messaging and contractual terms and conditions between ALM and individuals at the critical point of signup, this practice was found to be a contravention of the PIPEDA principle relating to the withdrawal of consent. The OPC has plainly said there is a high bar associated with charging a fee for deletion and, in order for such fees to be reasonable, further evaluation of factors such as the actual cost to the organization relative to the fee charged or the likely influence it would have on the individual's decision on whether to withdraw consent is required and, regardless, companies must communicate these fees prior to an individual providing consent.
As a result of this investigation, ALM (which has now changed its name to Ruby Corp.) and the OPC entered into a compliance agreement whereby ALM ultimately agreed to clean up its act in a number of areas, including taking stringent remedial measures regarding safeguards, retention, accuracy and transparency.
ALM clearly has a lot of homework to do as many of these actions are required to be fulfilled by the end of the year with others to be completed by May 31, 2017. It remains to be seen whether ALM will meet all of the requirements set out in the compliance agreement, but, in the meantime, prospective users of Ashley Madison may want to think twice about using ALM's various web sites. Buyer beware.
