FedEx’s July 2017 annual report disclosed a cyberattack it could not yet measure and had no insurance to cover
FedEx’s 10-K, filed July 17, 2017, said an attack carried by a Ukrainian tax software product had hit TNT Express. In our reading, it set out four unknowns, each owned by a different function, that every organization should plan for.
Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.
Two-minute summaryThe dek and the first sentence of each section, quoted from the article
FedEx’s 10-K, filed July 17, 2017, said an attack carried by a Ukrainian tax software product had hit TNT Express. In our reading, it set out four unknowns, each owned by a different function, that every organization should plan for.
- The week On July 17, 2017, FedEx Corporation filed its annual report on Form 10-K for the fiscal year ended May 31, 2017.
- The entry point: an attack carried by a supplier’s software The filing’s first explanation was causal.
- The recovery: when, whether and how much data The second unknown was operational.
- The money: a material loss without a number, and without insurance The third unknown was financial.
- The controls: an attack on the systems that produce the numbers The fourth unknown drew the least attention and, for a public company, carried heavy consequences.
- How the four unknowns reach banks, health systems and agencies FedEx is a logistics company.
- Objections, in order Our reading has reasonable critics.
- What a disclosure committee could check The FedEx filing suggests questions a disclosure committee, or its equivalent in a nonpublic organization, could ask before an incident.
- With hindsight FedEx’s next quarterly report, filed in September 2017, answered several of the July unknowns.
The week
On July 17, 2017, FedEx Corporation filed its annual report on Form 10-K for the fiscal year ended May 31, 2017. A short new section, under the heading TNT Express Cyber-Attack, described something that had happened after the year closed. [1]
FedEx had acquired TNT Express, an international express delivery company, on May 25, 2016. The filing described TNT as delivering to over 200 countries with more than one million shipments daily. [1] On June 28, 2017, FedEx had announced that TNT’s worldwide operations were significantly affected by a cyberattack known as Petya. [1]
The filing explained how the attack arrived. It involved the spread of an information technology virus through a Ukrainian tax software product. TNT operates in Ukraine and used the compromised software, which allowed the virus to infiltrate TNT systems and encrypt its data. [1] The systems and data of all other FedEx companies were, the filing said, unaffected. [1]
The filing described the state of recovery as of its date. All TNT depots, hubs and facilities were operational and most TNT services were available. But customers were still experiencing widespread service delays, including in invoicing, and manual processes were being used for a significant portion of TNT operations and customer service. [1]
FedEx said it could not estimate when TNT services would be fully restored. Its teams were restoring remaining operational systems and then finance, back-office and secondary business systems. [1] It added that it was reasonably possible TNT would be unable to fully restore all affected systems and recover all of the critical business data the virus had encrypted. [1]
On the money, FedEx said it was still evaluating the financial impact, but that the impact was likely to be material. It said it did not have cyber or other insurance in place that covered the attack. [1] It reported lost revenue from decreased volumes at TNT and incremental costs for contingency plans and remediation, without quantifying them. [1]
A new risk factor listed the consequences that could follow. The list included lost revenue from the disruption, bad debt from the inability to invoice properly, permanent loss of customers, remediation costs, contingency costs, investment in enhanced systems and customer incentives. [1] It also named reputational damage, potential litigation or investigations, any data loss later discovered, the loss of critical business data, and a longer and more costly integration of TNT with FedEx Express. [1]
The list ended with an item that is unusual in a filing: other consequences of which the company was not currently aware but would discover through remediation. The filing also said the attack might materially affect FedEx’s disclosure controls and procedures and its internal control over financial reporting in future periods. [1]
The filing set the attack against an earlier one. In May 2017, it said, FedEx had been one of the companies attacked by the WannaCry ransomware, which had not caused a material disruption or material costs. [1] Petya was different. The filing also stated that no data breach or data loss to third parties was known to have occurred as of its date. [1]
Contingency plans used both networks. FedEx was moving TNT packages within the FedEx Express network and offering the full range of FedEx Express services to TNT customers as alternatives. [1] For readers in banks, health systems and agencies, the filing was a public account of a destructive attack still in progress, and of what management could and could not yet say about it. [1]
A reader of a cyber incident disclosure looks first for the number. The July 2017 FedEx filing had no number to give. What it gave instead was a list of what it did not know, stated plainly while recovery was still under way. [1]
Our argument is that the filing can be read as a map of four separate unknowns, each of which belongs to a different owner inside an organization. Treating them as one incident question, owned by security, misses three of them. [1] [6] [7] The four are:
- The entry point: an attack that arrived through a supplier’s software, in a unit acquired in May 2016.
- The recovery: whether systems would come back, when, and whether encrypted data could be recovered at all.
- The money: a likely material loss that could not be sized, with no insurance to absorb it.
- The controls: the possibility that the attack would weaken the company’s disclosure and financial reporting controls.
The first belongs to procurement and third-party risk. The second belongs to technology operations and the business units. The third belongs to finance and risk transfer. The fourth belongs to the controller and the disclosure committee. Together they cover the incident from its cause to its effect on what the company could reliably report. After the four parts, we look at how the four unknowns reach banks, health systems and agencies, take up objections, and list what a disclosure committee could check.
A note on sources and our reading. We describe FedEx’s position from its own filings, which state the company’s view. We describe the attack’s mechanics from the federal alert later updated to reflect the government’s attribution. Statements about what disclosure rules required are our reading of the SEC’s releases. Organizations should confirm their own obligations with counsel.
The entry point: an attack carried by a supplier’s software
The filing’s first explanation was causal. The virus spread through a Ukrainian tax software product, which TNT used because it operated in Ukraine. [1] FedEx did not say it had been targeted. It said it had used software that was compromised.
How the attack moved
The federal government’s alert on the malware, later revised, gave the mechanics. It said the delivery mechanism during the June 27, 2017 event was a Ukrainian tax accounting software package. The threat actors, it said, had used a backdoor to compromise the software’s development environment as far back as April 14, 2017. [4]
Once inside a network, the alert said, the malware spread by several methods. They included stolen Windows credentials used with legitimate administration tools, and the EternalBlue and EternalRomance exploits against unpatched systems running SMBv1. [4] Microsoft had released a security update for that vulnerability on March 14, 2017. [4]
The alert also said the malware behaved more like destructive malware than ransomware. Based on its encryption methods, recovery of files appeared unlikely even if a ransom were paid. [4] That finding explains FedEx’s caution about recovering encrypted data. Paying was not a recovery plan.
What patching would and would not have done
The alert’s account of spreading methods carries a practical point about patching. One path, the SMB exploits, depended on systems that lacked the March 2017 update. [4] The other path used stolen credentials and legitimate Windows tools, which a patch does not close. [4]
In our reading, that combination means patching would have narrowed the spread but not stopped it. An organization with every system patched would still have faced credential-based movement from the first infected machine. Limiting administrative credentials and separating networks address the second path; patching addresses the first. [4]
For a security leader, the point is to plan against both paths at once. A patch compliance report that reads 100 percent is a good result. It does not answer whether one compromised workstation can reach every server using credentials cached on it. [4]
A supplier’s update as the door
In our reading, the entry point is the part a standard third-party review would struggle to catch. The software was legitimate. It was used for a legitimate business purpose, tax compliance in one country. The compromise happened in the supplier’s development environment, upstream of anything the customer could inspect. [4]
A vendor questionnaire asks about a supplier’s security. It does not stop a trusted update from carrying malicious code. The controls that could have limited the damage sat inside the customer’s own network: segmentation, credential hygiene and patching of the SMB vulnerability the malware used to spread. [4]
Questions a buyer can ask a software supplier
No questionnaire would have revealed a backdoor its own supplier had not found. But a buyer can ask questions that shape exposure. They include how updates are signed, how the supplier would detect a compromise of its build environment, how quickly it would tell customers, and whether the customer can stage updates before they reach every machine.
These questions do not prevent a supply chain attack. They decide how far one spreads and how fast the customer hears about it. In our reading, the answers weigh heaviest for software that runs with broad privileges on many machines, as accounting and tax packages can. [4]
A recent acquisition
FedEx had acquired TNT in May 2016. [1] The attack came while the two were still being combined. The filing’s risk factors on integration listed the combining of administrative and information technology infrastructure and computer systems as an issue that had to be addressed. [1] In our reading, an acquired unit’s local software, used in one country for one function, is the kind of asset an integration program inventories late.
FedEx’s later quarterly report said the company was accelerating portions of the TNT integration given the cyberattack. [2] The attack changed the order of integration work. Systems that might have been retired later were addressed sooner.
Local software in global groups
TNT used the compromised package, the filing said, because it operated in Ukraine. [1] In our reading, a tax package of that kind exists to meet one country’s rules. A multinational group can run several such local tools: payroll for one country, tax filing for another, regulatory reporting for a third. Each can be chosen locally, supported locally and missed by group security.
In our reading, these tools deserve their own entry in the group’s software inventory. They can combine broad access, because they handle finance data, with narrow oversight, because they serve one jurisdiction. A bank with foreign branches, a health company with overseas research sites or an agency with offices abroad holds the same kind of asset.
Containment by structure
The filing said the systems and data of the other FedEx companies were unaffected. [1] It did not explain why. We do not speculate on FedEx’s architecture. We note only that the separation between TNT and the rest of the group, whatever its cause, defined the blast radius. For any group with recently acquired units, the boundary between networks is a resilience control as well as an integration milestone.
The recovery: when, whether and how much data
The second unknown was operational. The filing separated three questions that are easy to blur: when services would be fully restored, whether all affected systems could be restored at all, and whether all critical encrypted data could be recovered. [1]
Services before systems
By the filing date, depots, hubs and facilities were operational and most services were available. Customers still faced widespread delays, and manual processes carried a significant part of operations and customer service. [1] The business was running. Its systems were not.
That distinction matters for anyone setting recovery objectives. A business can restore service by moving work to people and to another network, as FedEx did with FedEx Express. [1] That restores revenue. It does not restore the records, the invoicing or the controls that run through the systems.
Recovery in order of priority
The filing described a sequence. Critical systems first. Then remaining operational systems, then finance, back-office and secondary business systems. [1] Invoicing delays appeared in the risk factor as a source of lost revenue and bad debt. [1]
In our reading, the sequence shows a trade-off in destructive incidents. Operations come back before finance because operations earn the revenue. But finance systems record it. A long gap between the two produces exactly the invoicing and bad-debt risk the filing named. [1]
The data question
One line in the recovery discussion carried the greatest weight: the possibility that TNT would be unable to recover all of the critical business data the virus encrypted. [1] The filing did not say whether backups existed or why they might not suffice. It said only that full recovery was not assured.
The later quarterly report resolved this. It said substantially all of TNT’s critical operational systems had been restored and critical business data had been recovered by the time of that filing in September. [2] The July filing’s caution turned out to be prudent rather than predictive. It was still the right thing to say on July 17, when the answer was not known. [1] [2]
Custom work comes back last
The September quarterly report described the last stage of recovery. Core shipping services were back. FedEx was focused on restoring key customer-specific specialized solutions and systems in time for the peak shipping season. [2] Its updated risk factor listed failure to restore those solutions in time as a possible source of lost revenue. [2]
In our reading, this stage is easy to overlook in recovery planning. Standard services run on standard systems, which are documented and rebuilt first. Bespoke integrations for large customers run on configurations that may exist in few places and in few people’s heads. They also serve the customers with the deepest commercial ties.
A recovery plan that ranks systems only by internal criticality can leave these integrations until last. A plan that also ranks them by the revenue and contracts they support may order the work differently. FedEx’s own disclosure shows the commercial stakes of that ordering. [2]
Manual work as a new risk
Manual processes kept TNT running. They also carried risks of their own, which the filings named indirectly: invoicing delays, bad debt, and service failures that could lead to claims. [1] [2] A business that runs manually for weeks accumulates records that later have to be reconciled with restored systems.
For a health system or a bank, manual work during an outage carries regulated risk as well as financial risk. Paper orders, manual payments and offline records each need a reconciliation plan before the outage, not after it. The FedEx filings show how long the manual phase can last when systems must be rebuilt. [1] [2]
Rebuild, not restore
The alert described what the malware did to each machine it reached. It encrypted files and, with administrator rights, the master boot record, making the computer unusable. [4] A machine in that state is not cleaned. It is rebuilt.
That shapes the recovery plan. A ransomware plan built around decryption keys or file-level restores does not fit. The plan has to assume that every affected machine needs a fresh image and that data must come from copies the attacker could not reach. [4] The FedEx filing’s uncertainty about data recovery is consistent with that kind of work still under way. [1]
For a hospital with thousands of workstations and devices, or an agency with field offices, the rebuild rate becomes the recovery rate. In our reading, the planning number to know is how many machines the organization can reimage in a day, with what staff, from which trusted images.
Recovery as a disclosure input
The recovery state fed every other unknown. The financial impact depended on how long manual processes lasted. The controls question depended on when the systems that support financial reporting came back. [1] In our reading, a disclosure team cannot draft without a recovery status it trusts, dated and owned by someone in operations.
The money: a material loss without a number, and without insurance
The third unknown was financial. The filing said the impact was likely to be material, and that it could not yet be quantified. [1] That combination is uncomfortable for a public company and informative for everyone else.
Likely material, not yet measured
FedEx said it was still evaluating the financial impact given the recent timing and magnitude of the attack, and its initial focus on restoring operations and customer service. [1] It identified the categories of cost and lost revenue in its risk factor, without amounts. [1]
The SEC’s 2018 interpretive guidance later addressed this situation in general terms. It said a company may need time to discern the implications of an incident, and that an ongoing investigation would not on its own provide a basis for avoiding disclosure of a material incident. [6] In our reading, the FedEx filing had already taken that course: it disclosed likely materiality before it had a number. [1] [6]
Materiality before measurement
The order of the filing’s statements matters. It said the impact was likely material before it had an estimate. [1] A company can reach that judgment from the shape of the loss: a unit’s worldwide operations disrupted, services delayed and manual work in place. The number confirms the judgment later; it does not have to precede it. [1] [2]
In effect, the 2023 rules made that sequence routine for registrants. Item 1.05 requires a filing within four business days of determining materiality, and an amendment for information not determined or unavailable at first filing. [7] In our reading, a disclosure committee should practice reaching a materiality judgment from qualitative facts, because the quantitative ones may arrive months later.
No cover
The sentence on insurance was short. FedEx did not have cyber or other insurance in place that covered the attack. [1] It repeated the statement in its September quarterly report. [2] The filing did not say whether FedEx had considered such cover or why it was not in place. We do not guess.
For a risk committee, the line shows that the transfer question needs an explicit answer. A company can choose to retain cyber risk. It should know it has made that choice before an incident, and be able to say so in a filing afterward. The filing showed FedEx could state its position clearly. [1]
What insurance would and would not have changed
Insurance would have changed who bore part of the loss. It would not have changed the operational recovery, the invoicing delays or the customer losses the risk factor listed. [1] In our reading, the insurance question is real but secondary. The primary financial control is the speed of recovery, because lost revenue was the main category the later filings named. [2] [3]
Those later filings put figures on the loss. The September 2017 quarterly report estimated that the attack reduced first-quarter results by about $300 million, or $0.79 per diluted share, mainly from lost revenue and restoration costs. [2] The 2018 annual report estimated the impact at $400 million, or $1.19 per diluted share, during the first half of the fiscal year. [3]
How the loss was measured
The September estimate described its own composition. The $300 million impact came primarily from lost revenue due to decreased shipments in the TNT network, plus incremental costs to restore systems. [2] The figure is an estimate of lost earnings as well as a tally of spending. [2]
That matters for anyone comparing incidents. A loss figure built chiefly from lost revenue depends on a counterfactual: what TNT would have shipped without the attack. A figure built from invoices for recovery work does not. In our reading, a board reading an incident cost should ask which kind of number it is looking at.
The integration plan absorbed the shock
The September report tied the attack to the TNT integration. FedEx said integration expenses and capital spending would be higher than earlier estimates, because it was accelerating portions of the integration given the attack. [2] Its operating income target for the FedEx Express segment assumed continued recovery from the attack. [2]
In our reading, this is a cost of a destructive incident that loss tallies can miss. The company spent on recovery and also changed the timing of a larger program. Some of that spending would have come eventually. The attack moved it forward and made it more urgent.
Invoicing as a financial control
The risk factor named increased bad debt expense due to the inability to invoice properly as a consequence of the attack. [1] The September update restated it as bad debt from delayed invoicing. [2] Invoicing is where an operational outage becomes a balance sheet problem.
Customers who receive late or inaccurate invoices pay late or dispute them. Some do not pay at all. A company that cannot invoice for weeks also cannot see its own receivables clearly. In our reading, the finance function’s recovery priority should include the billing system alongside the general ledger.
Costs that arrived later
The risk factor also named costs that would not appear immediately: customer incentives to restore confidence, investment in enhanced systems, a costlier integration and possible claims. [1] The September report added claims for service failures and a higher effective tax rate from reduced international earnings to its list. [2] In our reading, a destructive incident has a long financial tail, and the first filing can only describe its shape.
The controls: an attack on the systems that produce the numbers
The fourth unknown drew the least attention and, for a public company, carried heavy consequences. The filing said the attack might materially affect FedEx’s disclosure controls and procedures and its internal control over financial reporting in future periods. [1]
Why controls were at risk
Disclosure controls and financial reporting controls run on systems. When the systems that record transactions are encrypted, the controls embedded in them stop working. Manual processes replace them, and manual processes have their own control gaps. [1] [2]
The September quarterly report described what happened. TNT’s internal control environment changed during the quarter, as controls and procedures were modified so they could be performed. By the end of the quarter, the report said, TNT’s control environment was functioning at pre-attack levels. [2] Management concluded that disclosure controls were effective as of August 31, 2017. [2]
Reporting on a unit whose controls were impaired
FedEx’s management had to evaluate group-wide disclosure controls while one unit operated with modified procedures. [2] The September report described the approach: it disclosed the change in TNT’s control environment during the quarter, stated that it had returned to pre-attack levels by quarter’s end, and then gave its effectiveness conclusion. [2]
In our reading, that sequence is a useful template. State what changed, state when it was restored, then conclude. A conclusion without the first two steps would ask readers to trust a judgment they could not test.
What the SEC later said about controls
The SEC’s 2018 guidance tied disclosure controls to cyber risk directly. It said disclosure controls should provide a way to discern the impact of incidents and a protocol to assess their potential materiality. [6] It also reminded companies of possible duties to correct or update earlier statements that became inaccurate. [6]
In our reading, the FedEx filing shows the controls question from the inside. The attack threatened both the systems that produce financial data and the process that decides what to disclose about the attack. The company had to report on its controls using information from a unit whose controls were impaired. [1] [2]
Selective disclosure
The 2018 guidance also reminded companies that Regulation FD can apply to cyber matters. When a company discloses material nonpublic information to certain persons, such as analysts, it must make that information public. [6]
During a long recovery, the pressure to brief investors, customers and partners privately is strong. Each briefing about recovery progress or expected costs could contain material information. In our reading, a company in FedEx’s July position needs one channel for anything material, and that channel is its public filings. [1] [6]
Insider trading and timing
The 2018 guidance also addressed trading. It encouraged companies to consider measures to prevent insiders from trading on undisclosed information about cyber incidents. [6] The FedEx filing does not discuss this, and we make no suggestion about anyone’s conduct. We note it because a period of known but unquantified loss is exactly when such measures matter. [6]
What the 2023 rules ask about suppliers
The SEC’s 2023 rules added a standing disclosure about processes. Item 106 of Regulation S-K asks registrants to describe their processes for assessing, identifying and managing material cybersecurity risks. [7] One listed item is whether the registrant has processes to oversee and identify risks associated with its use of any third-party service provider. [7]
Item 106 also asks for a description of the board’s oversight and management’s role. [7] In our reading, a company describing these processes after 2023 is, in effect, answering in advance the entry-point and controls questions the FedEx filing answered after the fact.
How the four unknowns reach banks, health systems and agencies
FedEx is a logistics company. The four unknowns it described apply, with different rules, to organizations in each of our sectors.
Banks and bank holding companies
A bank holding company with registered securities faces the same SEC disclosure framework FedEx did. [6] [7] A bank’s entry-point risk includes software used for one jurisdiction’s regulatory reporting, much as TNT’s tax software was. Its recovery question includes whether core banking and payment records can be restored intact. Banking regulators have their own incident rules, which we do not analyze here.
The money unknown is sharper for a bank than for a shipper. A bank’s lost revenue during an outage is joined by customer harm: payments not made, balances not visible. The filing’s list of consequences would, for a bank, need regulatory and customer remediation lines added. [1]
Credit unions and community banks
Institutions without registered securities do not file under the SEC’s cyber disclosure rules. The four unknowns still apply. A credit union’s board needs answers to the same questions about entry points, recovery, insurance and controls, even if no public filing follows.
Health systems
A health system that is not an SEC registrant has no Form 8-K duty; a publicly traded health company files like FedEx. The recovery unknown has a clinical dimension that a logistics company lacks: manual processes in patient care carry safety risk as well as cost. [1]
The invoicing problem FedEx described has a direct parallel in revenue cycle systems. A health system that restores clinical systems before billing systems will carry the same bad-debt and delayed-cash risk FedEx named. [1]
Public agencies
Agencies do not report to investors. They report to legislatures, oversight bodies and the public. The four unknowns map onto those reports. In particular, an agency whose finance systems are encrypted may face questions from auditors about its financial statements, the public-sector version of the controls unknown. [1]
The entry-point unknown also applies to agencies. An agency that uses a commercial package for one program, such as benefits payments or licensing, depends on that vendor’s build environment as much as TNT depended on its tax software supplier. [4] The recovery unknown applies with public consequences: citizens waiting for services, and staff working on paper. In our reading, an agency’s continuity plan should name which services it would run manually, for how long, and how it would reconcile the records afterward.
Risk managers and insurance buyers
The FedEx filing gave risk managers a public example of a large company stating that it had no insurance covering a cyberattack. [1] Whatever the reasons, the statement made the retention decision visible to investors. A risk manager in any sector can ask whether the organization’s board knows its own position as clearly.
The later loss figures also show what insurance could and could not reach. The September report said the impact came primarily from lost revenue due to decreased shipments. [2] Business interruption cover addresses that kind of loss in principle, subject to policy terms we do not analyze. Restoration costs are a separate category with different terms.
Suppliers and service providers
Every organization in these sectors is also a customer of companies like FedEx. TNT’s customers experienced delays and manual service. [1] A bank or hospital that depended on a disrupted carrier for time-sensitive shipments was, in effect, inside FedEx’s incident without being a party to it. In our reading, that argues for contingency plans that assume a key supplier is unavailable for weeks.
Objections, in order
Our reading has reasonable critics. We take the strongest objections in turn.
1. The filing was ordinary risk-factor boilerplate
Risk factors can read as boilerplate. This one did not. It described a specific attack, a specific entry point, a specific absence of insurance and a specific possibility of unrecoverable data. [1] The specificity is what makes it useful.
2. The four unknowns are one incident and should have one owner
Incident response does need a single commander. But the questions the filing raised are answered by different functions: procurement, operations, finance and the controller. [1] One commander coordinating four owners is different from one owner answering four questions alone.
3. FedEx was unlucky, and the lesson is about one supplier
The federal alert described a backdoor planted in a supplier’s development environment months before the attack. [4] That technique is not specific to one supplier. Any organization that accepts updates from software vendors carries the same structural risk.
4. Insurance would have solved the money problem
Insurance would have shifted part of the loss. It would not have prevented the lost revenue from disruption and customer loss, which the later filings identified as the main cost. [2] [3] The primary control was recovery speed.
5. Today’s rules make this history
The SEC adopted incident disclosure rules in 2023, and they changed timing and form. [7] They did not change the four unknowns. A company filing under the new rules still has to say what it knows and does not know about entry, recovery, money and controls. [7]
6. Saying what you do not know invites litigation
Disclosing uncertainty has legal risk. So does omitting it. The SEC’s 2018 guidance said an ongoing investigation would not on its own justify avoiding disclosure of a material incident, and reminded companies of duties to correct and update. [6] In our reading, a clear statement of what is unknown, updated as facts arrive, is the more defensible course, and it is the course FedEx took. [1] [2]
7. A state-sponsored attack is beyond any company’s control
The White House later attributed the attack to the Russian military. [5] A company cannot be expected to stop a determined state actor at the point of entry, and we do not suggest FedEx could have. [5]
But the alert showed that, once inside, the malware spread through ordinary weaknesses: reusable credentials, legitimate administration tools and an unpatched protocol. [4] The attacker’s identity did not change those paths. In our reading, attribution answers who did it. It does not answer how far it could go inside a given network, which remains a matter of the defender’s own controls.
What a disclosure committee could check
The FedEx filing suggests questions a disclosure committee, or its equivalent in a nonpublic organization, could ask before an incident.
- Which software in our environment is used for one country’s or one regulator’s requirements, and who owns its update path.
- What is our order of recovery, and when do finance and invoicing systems return relative to operations.
- Can we recover critical encrypted data from backups the attacker could not reach, and when did we last prove it.
- Is our cyber risk transfer position a recorded decision, and can we state it in one sentence.
- Which financial reporting controls depend on systems that could be encrypted, and what manual controls replace them.
- Who decides materiality when the number is not yet known, and on what recovery information.
These questions are not new. What the FedEx filing adds is evidence of why each needs an owner before the incident. The company had to answer all six in public, under time pressure, while its unit ran on manual processes. [1]
The first question addresses the entry point. The second and third address recovery. The fourth addresses the money. The fifth and sixth address controls. A committee that can answer all six has, in our reading, prepared for the filing FedEx had to write on July 17, 2017. [1]
With hindsight
FedEx’s next quarterly report, filed in September 2017, answered several of the July unknowns. It said substantially all TNT services had been fully restored during the quarter, and that substantially all critical operational systems had been restored and critical business data recovered. It added that TNT’s revenues, volumes and profits remained below pre-attack levels. [2] It estimated the first-quarter impact at about $300 million, and repeated that FedEx had no insurance covering the attack. [2]
The same report described the controls outcome. TNT’s control environment had changed during the quarter, as procedures were modified to be performed, and was functioning at pre-attack levels by the quarter’s end. [2] Management concluded that disclosure controls were effective as of August 31, 2017. [2]
The 2018 annual report closed the loop. It said all of TNT’s critical operational systems were fully restored, critical business data recovered and shipping services and solutions back in place as of the second quarter of the fiscal year. [3] It estimated the attack’s impact at $400 million, or $1.19 per diluted share, in the first half of the year. [3]
Attribution followed. On February 15, 2018, the White House said the Russian military had launched the attack in June 2017. Its statement called NotPetya the most destructive and costly cyberattack in history and said it had caused billions of dollars in damage across Europe, Asia and the Americas. [5] The federal alert on the malware was updated the same day to reflect the attribution. [4]
Regulation followed too. Later in February 2018, the SEC issued interpretive guidance on public company cybersecurity disclosures. It addressed materiality, the timing of disclosure during investigations, duties to correct and update, disclosure controls and insider trading. [6] Each topic mapped onto a question the FedEx filing had faced. [1] [6]
In July 2023, the SEC adopted rules requiring a Form 8-K under new Item 1.05 within four business days after a company determines that a cybersecurity incident is material. Companies other than smaller reporting companies had to comply from December 18, 2023. [7] The rules also require companies to amend the filing to add information that was not determined or was unavailable when they first filed. [7]
That amendment requirement is, in our reading, the clearest echo of July 2017. FedEx had disclosed materiality and an unknown amount, then supplied the amount in later reports. [1] [2] The 2023 rule formalized that sequence for every registrant. The SEC’s release also cited growing reliance on third-party providers for information technology among the reasons for its rules. [7]
What held up from the July filing? Its caution held. It said full recovery of data was uncertain, and recovery succeeded, but no one could have known that on July 17. [1] [2] Its materiality judgment held: the later filings confirmed a material loss. [2] [3] Its statement on insurance held, as a fact about FedEx’s position. [2]
One more test of the July filing is its list of possible consequences. Several appeared in later filings as real costs: lost revenue, restoration costs and a faster, costlier integration. [1] [2] The September update added two potential consequences the July list had not named: claims for service failures, and a higher effective tax rate from reduced international earnings. [2] In our reading, the July list was a fair forecast, and it was incomplete, as its own last item admitted. [1]
Nor was the attack FedEx’s alone. The White House statement described the same attack causing damage across three continents. [5] The structural risk the filing exposed, a trusted supplier’s update carrying destructive code, remains, in our view, a central concern for third-party risk programs in all three of our sectors.
Spotted an error? Tell us and we will check it and correct the article where needed.