Information Governance Think Tank
The UK’s Information Commissioner, John Edwards, has issued a reprimand to the Department for Education following the prolonged misuse of the personal information of up to 28 million children and a failure to do due diligence on who could access pupils’ learning records. An employment screening firm, trading as Trustopia, used the database to assist another organisation in checking if people opening online gambling accounts were 18.
Read more here.
In this article from the World Economic Forum, Tom Jackson and Ian R. Hodgkinson identify that organisations need to think about how to manage their data to minimise their digital carbon footprint.
Storage of ‘dark data’ defined as single-use data in the article, data takes up space on servers and results in increased electricity consumption. The authors point out that digitization generated 4% of global greenhouse gas emissions in 2020.
To read more on how dark data contributes to carbon emissions, and how organisations can lower their carbon footprint, click here.
The Notifiable Data Breaches (NDB) scheme commenced in February 2018, introducing new obligations for Australian government agencies and private sector organisations with an annual turnover of $3 million AUD or more. Notably, under the NDB scheme organisations are required undertake an assessment should they suspect:
If the assessment indicates that serious harm is likely to result from a data breach, they must notify the Office of the Australian Information Commissioner (OAIC) as well as all affected individuals so they can take action to address possible consequences and also. As data breaches and subsequent investigations are often significantly complex, an organisation or agency is given a baseline of 30 days to assess whether a data breach is likely to result in serious harm. However, once the organisation has formed the view that a data breach has occurred, individuals who may be seriously impacted by the data breach must be notified as soon as practicable. For example, in their recent data breach Optus has indicated that the assessment process took place over the course of no more than a couple of days prior to start of the notification process.
The OAIC has published bi-annual reports summarising the details of reported data breaches since 2018 and this article examines some of the identifiable trends in these reports over the past four years. The OAIC report for the most recent 6-month period (Jan-Jun 2022) should be released in the next few weeks, however some released statistics from the impending report indicate that the observed trends discussed in this article continue through the most recent period. The full OAIC reports are available from https://www.oaic.gov.au/privacy/notifiable-data-breaches/notifiable-data-breaches-statistics, and further information is available on the OAIC website: https://www.oaic.gov.au/privacy/data-breaches.
Organisations that fail to report a notifiable data breach can be subject to the same penalty as if they committed a serious or repeated breach of privacy, however organisations may look for a defensible reason to avoid reporting a breach as opposed to having to report a data security failure to the regulator. Even serious breaches where substantive personal data has been lost might be considered non-notifiable should the breached organisation feel they have undertaken sufficient remediation action which lessens the chance of serious harm.
Ransomware attacks, one of the common externally perpetrated data breach events, have evolved in recent years to extend beyond holding data in an encrypted state and often now include the exfiltration of sensitive data from target organisations. Data is often held by the attacker, with the threat of publication on the dark web balanced against payment of the ransom. Should the breached organisation make payment of the ransom, the attacker generally will agree to delete the exfiltrated data. While it is essentially impossible to be certain that all copies of this data have been deleted, a breach organisation is likely to consider such an arrangement as sufficient “remediation” of the breach event in that the likelihood of the data being used in such a way that would cause serious harm to individuals is substantially reduced. In such circumstances, the breached organisation may choose not to report the incident to the OAIC, despite the severity of the initial data breach.
Due to the complex nature of data breaches and reclassification of notifications over time, there is some variation in breach notification statistics between the time of OAIC publication and the present date. The stats shown in this article are taken from each quarterly/bi-monthly report which reflected notification data at the time of report publication, however it should be noted that breach statistics will have changed to a degree from what we have summarised from the OAIC reports.
Over the past 4 years, there have been more than 3,500 reported data breaches, of which 60% were Malicious (or criminal), 35% were Human Error and 4% were due to System Faults. Taking into account that 2018 was a partial reporting year, approximately 1,000 breaches would have been reported to the OAIC across 2018 and 2019, with a slight uptick in 2020 and a substantial reduction in 2021. A summary of the data breach notifications made to the OAIC are displayed in the table immediately below
| Year | Breaches | Change | Malicious | Change | Human error | Change | System fault | Change |
| 20181 | 749 | 449 | 265 | 35 | ||||
| 2019 | 997 | 0% | 625 | +4% | 329 | -7% | 43 | -9% |
| 2020 | 1,057 | +6% | 627 | 0% | 380 | +16% | 50 | +16% |
| 2021 | 910 | -14% | 545 | -13% | 324 | -15% | 41 | -18% |
| Total | 3,713 | 2,246 | 1,298 | 169 |
The table above also shows that malicious action breach incidents (combining both internal and externally originated) increased in 2019 compared against “non-malicious” breach types, however this trend was reversed in the following year. This was followed by was a slight reduction in all types of breach notifications in 2021. Overall, the OAIC NDB reports show a fairly flat trajectory over the length of the scheme.
Conversely, US data on publicly reported data breaches over this time period shows year on year increases between 2018 and 2021, effectively doubling from 1,244 to 2,407 per year over this time period2. It is worth noting that data breach reporting requirements in the US vary from state to state and are substantially different to the Australian NDB scheme. However, the significant difference in the overall trend of breach reports is still interesting, particularly as the US Data indicate that the number and severity of malicious cyber-attacks appear to be increasing over time 3. in contrast, the declining number of reported malicious breaches (as well as non-malicious breaches) in the Australian NDB data suggests there may be other systemic factors at play with regards to the number and type of breaches reported under NDB scheme.
As can be seen in the table below, the majority of data breaches tend to have low numbers of individuals affected by the breach. However, the number data breaches affecting large numbers of individuals remained fairly steady over the data period, indicating that a significant proportion of the Australian population is likely to have been impacted by a data breach throughout this time period.
| Year | Total Breaches | <1k | 1k-5k | 5k-10k | 10k-25k | 25k-50k | 50k-100k | 100k-250k | 250k-500k | 500k-1m | >1m | Unknown |
| 2018 | 749 | 638 | 61 | 15 | 9 | 1 | 4 | 4 | 1 | 0 | 2 | 14 |
| 2019 | 997 | 834 | 74 | 19 | 16 | 7 | 5 | 5 | 2 | 0 | 5 | 30 |
| 2020 | 1,057 | 890 | 90 | 12 | 13 | 14 | 6 | 2 | 4 | 2 | 7 | 13 |
| 2021 | 910 | 787 | 71 | 12 | 15 | 7 | 5 | 4 | 2 | 1 | 4 | 2 |
| Total | 3,692 | 3,149 | 296 | 58 | 53 | 29 | 20 | 15 | 9 | 6 | 16 | 59 |
The majority of personal information present in data breaches was contact information, followed almost equally by financial and identity information. The proportions of personal information types present in data breaches has not changed significantly year to year, indicating that there has not been significant changes in the how organisations are holding or protecting particular types of personal information over this period.
The higher number of breaches relating to contact information will be, to some degree, a function of the fact that only certain organisations need to hold more specific personal information about their customers. Despite the apparent lower impact from breaches relating to contact information, such data is still of substantial value for cyber-criminals for use in phishing and other targeted attacks, and may also be combined with information from other data breaches for more specific criminal use.
| Year | Contact Info | Financial Details | Identity Information | TFN | Health | Other sensitive information |
| 2018 | 647 | 335 | 273 | 186 | 148 | 61 |
| 2019 | 817 | 398 | 293 | 255 | 157 | 85 |
| 2020 | 890 | 408 | 439 | 272 | 184 | 134 |
| 2021 | 803 | 376 | 432 | 256 | 184 | 140 |
| Total | 3,157 | 1,517 | 1,437 | 969 | 673 | 420 |
Most forms of malicious/criminal attack have been fairly consistent year-on-year, however ransomware in particular has been increasing year on year and 45% of all ransomware incidents occurred in 2021. The steady increase in this form of money-motivated cyber-attack aligns with anecdotal and industry reports of increases in this type of activity from organised cyber-criminal gangs and certain nation-state actors.
It is interesting to consider the Optus data breach in the light of whether it would be considered a maliciously originated breach, a system fault, or a combination of both. While it certainly appears the case that a maliciously motivated individual or group has exfiltrated Optus customer data, the methods used remain a matter of debate and have not been fully confirmed by Optus or the federal government. Should it have been the case, an oversight which results in an API (Application Programming Interface) connected to a customer details database being left in an open state to external connections would almost certainly be viewed as a failure of internal systems and procedures. Optus’ CEO has indicated that the breach cause was not as straightforward as this, suggesting a more complex cause involving specific malicious technical action.
| Year | 2018 | 2019 | 2020 | 2021 | Total |
| Malicious Breaches | 449 | 625 | 627 | 545 | 2,246 |
| Theft paper/Storage | 73 | 80 | 53 | 61 | 267 |
| Social Engineering | 28 | 52 | 84 | 65 | 229 |
| Rogue Employee/Insider Threat | 41 | 71 | 60 | 54 | 226 |
| Cyber (ALL) | 307 | 422 | 430 | 365 | 1,524 |
| Cyber – Phishing | 125 | 146 | 132 | 113 | 517 |
| Cyber – Stolen Credentials | 79 | 140 | 108 | 100 | 426 |
| Cyber – Ransomware | 18 | 29 | 69 | 86 | 202 |
| Cyber – Hacking | 27 | 34 | 59 | 31 | 151 |
| Cyber – Brute | 34 | 25 | 30 | 18 | 107 |
| Cyber – Malware | 20 | 37 | 24 | 16 | 95 |
| Cyber – Other | 5 | 10 | 9 | 2 | 26 |
The majority of human error breaches are due to wrongly addressed emails, and this has been consistently the highest category, even with the 15% reduction in 2021. Unlike malicious or systemic breaches, human error breaches – as classified by the OAIC – have limited technical controls that can be implemented to assist with prevention. Instead, education and procedure remain the best defence against these type of breaches.
| Year | 2018 | 2019 | 2020 | 2021 | Total |
| Human Error Breaches | 265 | 329 | 380 | 324 | 1,298 |
| Wrong email recipient | 74 | 101 | 160 | 136 | 471 |
| Wrong hardcopy recipient | 33 | 30 | 37 | 18 | 118 |
| Loss of hard/soft storage | 34 | 40 | 25 | 23 | 122 |
| Unintended release/publication | 41 | 76 | 62 | 71 | 250 |
| Failure to use BCC | 22 | 18 | 30 | 25 | 95 |
| Failure to redact | 14 | 19 | 20 | 23 | 76 |
| Unauthorised verbal disclosure | 8 | 19 | 18 | 11 | 56 |
| Insecure disposal | 8 | 5 | 2 | 0 | 15 |
| Wrong Recipient (Other) | 19 | 21 | 25 | 17 | 82 |
| Other | 12 | 0 | 1 | 0 | 13 |
The Health, Finance and Business Services sectors collectively made up over 45% of all reported breaches in 2019 and 2020. In 2021, where a substantial reduction of breaches reports were made compared to previous periods, the combined breaches in there three sectors were still approximately 40% of the overall reports. Maliciously originated data breaches in the Legal, Accounting and Management sector was the only category to see a substantial rise in 2021, with almost all other types of breaches in these sectors seeing a decline from the previous year.
Given the level of highly personal information held by Health sector organisations, the fact that these organisations feature so highly in the NDB statistics is of cause for specific concern. While federal and state legislation provides guidance for the collection, management and use of health data, as well as highlighting the highly confidential nature of such data, Australia currently does not have an equivalent to the US HIPA Act where substantive penalties and sanctions can be levied specifically pertaining to non-criminal use or loss of health data.
| Sector | 2019 | 2020 | 2021 | Total |
| Health – Malicious | 111 | 97 | 87 | 295 |
| Health – Human Error | 106 | 135 | 74 | 315 |
| Health – System Error | 5 | 6 | 7 | 18 |
| Finance – Malicious | 77 | 98 | 57 | 232 |
| Finance – Human Error | 59 | 47 | 44 | 150 |
| Finance – System Error | 10 | 11 | 12 | 33 |
| Legal, accounting & management – Malicious | 60 | 42 | 61 | 163 |
| Legal, accounting & management – Human Error | 25 | 22 | 24 | 71 |
| Legal, accounting & management – System Error | 2 | 5 | 1 | 8 |
| Total | 455 | 463 | 367 | 1285 |
The OAIC provide details on the top 5 sectors reporting data breaches over each period. As only the Health, Finance and Legal, Accounting & Management sectors have consistently appeared in the periodic reports, and only the 2019-2021 reporting periods include complete data, only the data from those three sectors and three periods has been included in this analysis.
The OAIC official data breach statistics show an overall declining trend in reported breaches under the NDB scheme. On the surface this would potentially represent a good new story – in that organisations are becoming better at preventing data breaches and successful malicious attacks on organisations may becoming fewer. The counter-argument to this observation is the legal advice and remediation response organisations are using to inform their decisions on whether a breach falls under the NDB may have changed over time, resulting in fewer breaches being reported rather than fewer breaches actually occurring.
The recent data breach incident at Optus has highlighted the widescale impact that a large data breach can have both on the breached organisation and the individuals to which the data belonged to. In terms of the scale, size and type of data that was taken, in addition to the media coverage, there would be little chance that any person assessing this breach would consider that it would not require mandatory reporting. However, in circumstances where a less comprehensive data set was exposed, with substantially fewer affected individuals, the potential for serious harm may not be considered as high, resulting in variable decisions to report.
A smaller scope breach just involving loss of customer name and address information might be considered to hold lesser chance to cause serious harm by themselves. When such a breach is potentially remediated – say by payment of a ransom – it may be the case that an organisation feels that the breach no longer meets the threshold to require mandatory reporting and notification of affected individuals. However there is little in the way of guarantees that organisations can seek from cyber-criminals who hold exfiltrated data at ransom. The destruction of this data upon payment of a ransom is entirely in the control of the criminals and cannot be verified by the organisation.
It is also worth noting that a somewhat “lesser” data breach containing names and addresses may be combined with data sets containing account details, passwords and identify information obtained from other breaches. In a similar way that de-identified “Big Data” sets hold the potential for “re-identifcation” of individuals, combining multiple data sets residing on the dark web following successive breaches of different organisations results in a substantially higher chance of serious harm to affected individuals over time. As such, the OAIC and the Federal Government may wish to consider the provision of further guidance around notification requirements based on the type of data that exposed during a breach as well as what successful remediation of a breach should cover.
The Optus data breach has also demonstrated that certain types of organisation are required for regulatory reasons to collect more personal information than others. There appears to be substantial uncertainty in the various regulations governing this requirement as to the length of time such information needs to be held and also as to how such information can be used and must be protected by the collecting organisation. Undoubtedly both Federal and State governments in Australia have observed this issue in recent weeks and it can be hoped that specific actions clarifying and improving regulatory requirements around collection, storage, use and disposal of personal information by Australian organisations will be forthcoming in the near future.
Organisations may have other obligations outside of those contained in the Privacy Act that relate to personal information protection when responding to a data breach. These may include data protection obligations under state-based or international data protection laws. Notably, Australian businesses may need to comply with the European Union’s (EU’s) General Data Protection Regulation (GDPR) if they have an establishment in the EU, if they offer goods and services in the EU, or if they monitor the behaviour of individuals in the EU.
For data breaches affecting certain categories of information, other mandatory or voluntary reporting schemes may exist. For example, entities may need to consider reporting breaches to:
Office of the Australian Information Commissioner https://www.oaic.gov.au/privacy
Australian Cyber Security Centre (ACSC) https://www.cyber.gov.au/
IDCare https://www.idcare.org/about-idcare/what-is-idcare
Scamwatch https://www.scamwatch.gov.au/
MoneySmart https://www.moneysmart.gov.au/
The Office of the eSafety Commissioner https://www.esafety.gov.au/
Dr Peter Chapman, Director, Forensic Technology – KPMG
Thank you to Matthew Golab, Director – Legal Informatics and R&D at Gilbert and Tobin, for his analysis of the OAIC reports and contribution to this article.
Key Takeaways:
The Two Types of Data Your Organization is Accumulating (and Why You Should Care)
We’ll start with why you should care.
If you’re familiar with the data compliance space, you already know that new laws require your organization to take specific steps to protect the rights of anyone whose data they hold. (If you’re not familiar with data compliance – surprise!)
The first step to maintaining compliance with these laws is understanding what data your organization actually has. Not having this understanding is dangerous for three reasons:
The first step to understanding your data holdings is to understand the difference between the two main types of data: structured and unstructured.
Structured data is what probably comes to mind when you hear the word “data”: spreadsheets on spreadsheets filled with quantitative information. Essentially, structured data is anything you can store in rows and columns, such as information stored in databases (think SQL), CSV files, and so on. It’s easily understood and analyzed by applications other than the ones that generated it, and it doesn’t scale easily – which is good for privacy purposes. It doesn’t grow out of control on its own, at least not for a long time.
Unstructured data is the qualitative data naturally generated from interactions with people. Think the text stored in your emails, Teams chats, social media, and websites. It can also comprise images, PDFs, Word docs – anything you can’t store in rows and columns. It’s not usually in a format that other applications can easily understand and analyze. And it multiplies like you wouldn’t believe: how many emails have you sent and received just this week?
Both types carry their own risk, but unstructured data is by far the riskier of the two. In today’s world, we generate it so quickly and in such high volume – and with such little organization – that it’s functionally impossible to keep track of without using data discovery software.
On the flip side, once you have the right tool, getting to compliance becomes exponentially easier. When you can visualize all your unstructured data, you can see what’s out of compliance, fix that right quick, and understand where your policies and workflows need to change to keep everything above board.
Some Common Misconceptions Your Organization Might Have
We all know an ounce of prevention is worth a pound of cure. And yet: most of us don’t go to the doctor until we get sick.
The compliance world is no different. With regulations still relatively new, most organizations don’t fully appreciate the urgency surrounding the issue – and won’t until they themselves get breached.
If your organization is anything like most, you’re probably operating under one of the following misconceptions.
Misconception #1: We Already Know What Data We Have
Name the last time you checked your Teams log. Or your Downloads folder. Your email archives? You get the idea.
People – and companies – don’t typically monitor or clean these types of things without a push. Without the proper privacy functions in place, we’re liable to think the trash in the ocean isn’t a problem. Until, of course, there’s an island of it.
Misconception #2: We Won’t Get Breached
There is a roughly 30% chance your organization will get breached this year. This stat increases every year.
It’s also possible you’ve already been breached. According to IBM’s annual Cost of a Data Breach report, the average time to identify and contain a breach in 2021 was 287 days.
When you get breached, you can cut the time and expense involved significantly – nearly entirely – by already being in compliance. Compliant data equals a quick, cheap(er) remediation with no additional reputational damage on top of the fact that the breach occurred.
Misconception #3: It’s Too Expensive to Figure Out What We Have
According to that same IBM report, the average cost of a breach in a hybrid cloud environment was $3.61 million. On top of that, compliance failure was the top factor found to amplify data breach costs. And remember, it’s not just the cost of remediating compliance flaws you have to worry about. Regulatory fines are getting steeper every year.
Misconception #4: It’s Too Labor Intensive – We’d Need a Team of Experts
Since data privacy regulations are so new and the solution market is still growing, it’s easy to believe you’d need in-house specialists to operate whichever data discovery solution you ended up going with.
Not if you choose the right one! Specifically, you want to make sure you select a solution that’s purpose-built for ease of use. From deployment to monitoring and at every stage in between, no expert knowledge should be required. Don’t go with a solution that’s been repurposed from another area of the market, such as data loss prevention or data access management.
Misconception #5: Traditional Data Inventory Methods Still Cut It
Back in the day, and still sometimes today, organizations would build data inventories through manual assessments and questionnaires: they’d basically ask their staff what data they thought the organization had.
In today’s world, with data accumulating and multiplying by the second, a manual static inventory won’t do the trick. It’s obsolete as soon as you create it.
To ensure continuous compliance, you need real-time visibility into your data.
Learn more
To learn more about data protection, security and compliance, listen to this podcast published on the Society of Corporate Compliance’s blog.
For more information on how to achieve cost effective and lightning speed visibility into your unstructured data so you can mitigate risk, check out ActiveNav Cloud.
Author
Simon Costello, VP – APAC, ActiveNav
