Uber’s hidden 2016 breach shows why the decision to report an incident should not sit with security alone
On November 21, 2017, Uber disclosed that intruders had taken data on 57 million users in 2016, and that it had told neither regulators nor drivers. Trial evidence later described the matter being kept inside security. We argue the reporting decision needs owners outside it.
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
On November 21, 2017, Uber disclosed that intruders had taken data on 57 million users in 2016, and that it had told neither regulators nor drivers. Trial evidence later described the matter being kept inside security. We argue the reporting decision needs owners outside it.
- The week On Tuesday, November 21, 2017, Uber published a statement on its newsroom titled 2016 Data Security Incident.
- The decision: classification inside one function Every incident response process includes a moment of classification.
- The payment: an extortion paid through a vulnerability program The second part concerns money.
- The obligations: notice duties, and the duty an open investigation creates The third part concerns what the company owed, and to whom.
- The accountability: the company, an individual and the record The fourth part concerns how responsibility was assigned.
- How the structure applies across sectors today Notice rules have multiplied since 2016, as the hindsight section describes.
- Objections, in order The argument that the reporting decision needs owners outside security faces objections.
- What a compliance officer could check A tabletop exercise is a cheap way to test the checks.
- With hindsight The FTC moved first.
The week
On Tuesday, November 21, 2017, Uber published a statement on its newsroom titled 2016 Data Security Incident. The statement said the company had recently learned that, in late 2016, it had become aware that two individuals outside the company had inappropriately accessed user data. The data was stored on a third-party cloud-based service the company used. The statement said the incident did not breach Uber’s corporate systems or infrastructure.
The statement described what was taken. The individuals downloaded files containing the names and driver’s license numbers of around 600,000 drivers in the United States. They also downloaded some personal information of 57 million Uber users around the world, including names, email addresses and mobile phone numbers. Outside forensics experts had not seen indications that trip location history, credit card numbers, bank account numbers, Social Security numbers or dates of birth were downloaded.
The statement described the company’s response at the time. Uber had taken immediate steps to secure the data and stop further unauthorized access. It had identified the individuals and obtained assurances that the downloaded data had been destroyed. It had also implemented security measures to restrict access to and strengthen controls on its cloud-based storage accounts.
The statement then addressed the delay. It acknowledged the question of why the company was only now speaking about an incident from the previous year. It said an investigation had examined what happened and how it was handled, including the failure to notify affected individuals or regulators the previous year. Two of the individuals who led the response to the incident were, effective that day, no longer with the company.
The statement listed actions. Uber was individually notifying the drivers whose license numbers were downloaded and offering them free credit monitoring and identity theft protection. It was notifying regulatory authorities. It was monitoring the affected accounts and had flagged them for additional fraud protection, while saying it had seen no evidence of fraud or misuse tied to the incident.
The disclosure followed a related event. On August 15, 2017, the Federal Trade Commission had announced a proposed settlement with Uber over its privacy and data security claims. The FTC alleged that Uber had failed to provide reasonable security for consumer data stored with a third-party cloud provider. It said an intruder had accessed personal information about Uber drivers in May 2014, including more than 100,000 names and driver’s license numbers.
The FTC’s August complaint described specific practices. It alleged that Uber allowed engineers to use a single access key with full administrative access to the data, rather than distinct keys, and did not require multifactor authentication. It alleged that Uber stored sensitive consumer information, including geolocation information, in plain readable text in cloud database backups. The proposed order required a comprehensive privacy program and independent audits every two years for 20 years.
The November disclosure therefore landed on top of an open matter. The August settlement was proposed, not final, and it addressed a 2014 incident. The 2016 incident had happened while the FTC was still investigating Uber’s data security, and it had not been disclosed to the Commission. How that came about, and what followed, became the subject of later FTC, state and federal criminal proceedings.
Breach coverage dwells on how attackers got in. The Uber disclosure of November 2017 is instructive for a different reason. The entry was ordinary for a cloud incident: an access key, a storage service and downloaded files. What made it consequential was what happened inside the company afterward, when a small group decided what the incident was and who would hear about it.
Our argument is that the reporting decision was made, in effect, by the security function alone, and that this structure allowed concealment. Later records, including federal trial evidence described by prosecutors and the company’s own admissions, show the decision did not reach the lawyers handling the FTC inquiry. For banks, health systems, public companies and agencies, the lesson we draw is structural. The decision whether an incident is reportable needs owners outside security.
We take the case in four parts. Each is a separate failure point, and together they cover the path from incident to disclosure:
- The decision: who classified the 2016 incident, and what that classification controlled.
- The payment: how money moved to the intruders, and through which program.
- The obligations: which duties to notify existed, including the duty created by an open investigation.
- The accountability: how responsibility was later assigned to the company and to an individual.
After the four parts, we look at how the same structure applies across sectors today, at objections, and at what a compliance officer could check. A note on sources: we rely on Uber’s own statement, FTC releases, and Department of Justice releases describing a non-prosecution agreement and a jury verdict. Where the FTC alleged facts, we say so. Where facts come from trial evidence that led to a conviction, or from the company’s admissions, we say that instead. Where we interpret a duty, that is our reading.
We do not name the individuals involved. The former chief security officer was convicted at trial, as the Department of Justice announced. The point of this article is the structure that allowed the decision, which any organization can examine in itself.
The decision: classification inside one function
Every incident response process includes a moment of classification. Someone decides whether an event is a security incident, whether it involved personal data, and whether it triggers notice. That decision controls everything after it. If an event is classified as a resolved vulnerability report, nobody downstream is told to look for a notice duty.
What the records show about the 2016 classification
The Department of Justice’s release on the 2022 verdict describes trial evidence on this point. It says the intruders contacted the chief security officer by email on November 14, 2016, and demanded a large payment. It says his staff verified the theft of records on about 57 million users and 600,000 driver’s license numbers. It says the evidence showed he executed a scheme to prevent knowledge of the breach from reaching the FTC.
The release says the evidence showed the matter was kept within the security group, and that the chief security officer continued working with the lawyers handling the FTC inquiry without mentioning it. In the 2022 non-prosecution agreement, Uber admitted that its personnel failed to report the November 2016 breach to the FTC despite the pending investigation. We present the first as trial evidence that prosecutors described and that led to a guilty verdict, and the second as the company’s own admission.
The technical facts the group would have weighed
The technical path was not exotic. The FTC’s revised complaint alleged that the intruders used an access key an engineer had posted on a code-sharing website. The non-prosecution agreement release says, as agreed facts, that the intruders used stolen credentials to reach a private source code repository and obtain a private access key. With the key, they copied large quantities of data from cloud storage.
A reportability group hearing those facts would have had little room for doubt. Personal data had been copied by outsiders who were demanding money. Whatever the security team’s view of the cause, the notice analysis would have started that day. The technical facts were never the obstacle to disclosure. The obstacle was who heard them.
Why one function cannot be the only judge
A security team has deep knowledge of what happened technically. It also has its own exposure when an incident occurs. In Uber’s case, according to the release, the chief security officer had testified to the FTC about the company’s security practices ten days before the intruders made contact. A function judging its own failure has a conflict, whatever the individuals’ intentions.
The conflict is not unique to Uber. Any security leader whose program has just failed is asked to decide how serious the failure was. A process that lets that judgment go unreviewed relies on the leader’s integrity alone. Good processes do not depend on any one person being right, and incident classification is a place where that principle matters.
Who else should be in the room
In our reading, the reportability decision belongs to a small group with separate responsibilities. Legal counsel decides what the law requires. Compliance or privacy decides what regulators and individuals must be told and when. Security supplies the facts. For a public company, the disclosure committee decides materiality. For a regulated firm, the officer responsible for regulatory relationships knows which open matters an incident might touch.
This group need not slow response. Containment and investigation can proceed under security’s control. The group’s job is narrow: to decide, on the facts security provides, whether the event is reportable and to whom. It should meet within hours of a significant event, not after the security team has decided the matter is closed.
The group also needs authority. If its decision can be overruled by the security leader, or ignored, it is advice and not a control. Its charter should say that the decision whether an incident is reportable belongs to the group, and that no notice, payment or statement to a regulator about the incident goes out without its sign-off. Security keeps full control of the technical response. The group controls what the organization says, and to whom.
Defining the trigger for escalation
The practical difficulty is deciding which events reach the group. If every alert goes there, the group drowns. If the security team decides what to escalate, the original problem returns. One approach is to define triggers that require escalation regardless of security’s assessment: confirmed access to personal data, an extortion demand, contact from the attacker, or any event touching a system under regulatory review.
An extortion demand deserves its own trigger. By definition, it means an outside party claims to hold data and wants payment. That claim alone should start legal and compliance review. Uber learned of the 2016 incident through exactly such a demand, according to the trial evidence. A rule that escalates every extortion demand, if followed, would have put the event in front of people outside security on the first day.
What security owes the group
The group can only decide well if security tells it what security knows. The minimum is short: what systems were touched, what data was accessed or taken, how confident the team is in each finding, whether the attacker has made contact or demands, and what remains unknown. Security should give this in writing at the first meeting and update it as the investigation moves.
The written summary matters for a second reason. According to the Department of Justice release, when new management asked about the incident in fall 2017, the chief security officer deleted from a subordinate’s draft summary that the hack involved personally identifying information and a very large quantity of user data. A summary that goes directly from the investigating team to the reportability group, with a record of edits, is harder to distort.
A simple escalation matrix
An organization can write its triggers as a short matrix that security cannot override. Each trigger names who must be told and how fast. The matrix below is our illustration, not a regulatory requirement:
- Any extortion or ransom demand: legal, compliance and an executive sponsor, the same day.
- Confirmed or likely access to personal, health or financial data: legal and privacy, the same day.
- Any incident touching a system, product or topic under regulatory inquiry or audit: the lawyers handling that matter, the same day.
- Any proposed payment to an outside party connected to an incident: legal and finance, before payment.
- Any incident that might be material to a public company: the disclosure committee, at once.
The matrix does not decide the outcome. It decides who is in the room. That is the step the 2016 events skipped, on the trial evidence the Department of Justice described.
Evidence that the decision was made properly
A classification decision should leave a record. The record should say who decided, on what facts, applying which legal analysis, and what was done. If an examiner, regulator or court later asks why an incident was not reported, the record answers. A decision made informally inside one team, with no record outside it, cannot be defended or checked.
The payment: an extortion paid through a vulnerability program
The second part concerns money. Paying an intruder is a significant decision with legal, financial and reputational dimensions. How Uber’s payment was made, and what paperwork surrounded it, became central to the later proceedings.
What the FTC and the trial evidence described
The FTC’s revised complaint in April 2018 alleged that Uber paid the intruders $100,000 through its third-party bug bounty program. The FTC noted that such a program exists to reward people who responsibly disclose vulnerabilities, not those who exploit them to access personal information. The Department of Justice’s 2022 release described trial evidence that the payment was made in bitcoin in December 2016, in exchange for non-disclosure agreements.
The release says those agreements contained a false representation that the intruders had not taken or stored data. It says the company later identified the two intruders and had them sign new agreements in their true names. Uber’s November 2017 statement, by comparison, said only that the company identified the individuals and obtained assurances that the downloaded data had been destroyed.
Why the channel mattered
A bug bounty program is a legitimate tool. It pays researchers who find and report flaws. We think bounty payments are often routine, small relative to company budgets, and handled by security teams. Those features can make a bounty program a channel through which a payment to intruders moves without the scrutiny a large, unusual payment would attract.
The paperwork also mattered. A bounty payment comes with terms that treat the recipient as a researcher who found a flaw. In our reading, using that framework recast the event, on paper, as a vulnerability report rather than a data theft. That is the classification problem again, expressed in a payment record.
Agreements with intruders
The non-disclosure agreements deserve separate attention. An agreement signed by an intruder cannot make stolen data disappear. Its promises are only as good as the intruder’s word. Uber’s November 2017 statement relied on assurances that the data had been destroyed. The trial evidence, as the release describes it, showed the agreements also contained a false statement that no data was taken or stored.
An organization that receives such assurances should treat them as one weak piece of evidence, not as closure. The notice analysis should proceed as though the data may still exist, unless forensic evidence shows otherwise. In our reading, an intruder’s promise does not change whether data was acquired without authorization, which is the usual trigger for notice under breach laws.
Controls on incident-related payments
The structural response is to treat any payment connected to a security incident as a separate category. That includes ransom payments, payments to extortionists, and bounty awards connected to data access rather than flaw discovery. Each should require approval from legal and finance, a record of the reason, and checks under sanctions and other law. Organizations should ask counsel how payment rules apply to them.
The FTC’s 2018 order shows what a regulator wanted after the fact. Among other things, it required Uber to retain certain records related to bug bounty reports involving potential or actual unauthorized access to consumer data. In our reading, that requirement treats bounty records as evidence a regulator may need. Organizations can adopt that discipline without an order.
Sanctions, law enforcement and other victims
Payments to intruders raise further questions that only counsel can answer. They include sanctions screening, reporting to law enforcement, and the effect on any insurance claim. The Department of Justice release adds another consideration. It says the evidence showed the security chief knew the intruders were hacking and extorting other companies. A payment that keeps an extortion secret can leave other victims exposed.
Distinguishing research from intrusion
A bounty program needs a rule for telling research from intrusion. A researcher who finds a flaw and reports it without taking data fits the program. Someone who downloads data and then asks for money does not, whatever they call themselves. The rule should say that any report involving the taking of personal data leaves the bounty process and goes to incident response, with the reportability group informed.
The obligations: notice duties, and the duty an open investigation creates
The third part concerns what the company owed, and to whom. In our reading, duties to notify individuals came mainly from state breach notification laws in 2016. Uber’s 2017 statement acknowledged a failure to notify affected individuals or regulators. A second kind of duty arose from the FTC investigation that was then pending.
Notice to individuals
State breach laws, in our reading, generally require notice to residents when certain categories of personal data are acquired without authorization. In our reading, driver’s license numbers fall within the definitions of several state laws. The 600,000 drivers whose names and license numbers were downloaded are the group Uber notified in November 2017. Whether and when notice was owed under each state law is a legal question we do not answer here.
The pending investigation
The second duty is easy for a security team to overlook. When a regulator is investigating a company’s data security, its written questions can require the company to disclose other incidents. The non-prosecution agreement release says the FTC’s written questions required Uber to provide information about any unauthorized access to personal information. A new breach during that investigation was responsive information.
This duty runs through whoever answers the regulator. If the people answering do not know about a new incident, the company can give incomplete answers without intending to. If someone who knows keeps it from them, the answers become misleading. The 2022 verdict, as the Department of Justice described it, turned on that second situation: obstruction of FTC proceedings.
Discovery and determination
The notice rules that apply today start their clocks in two different ways. HIPAA’s clock starts at discovery, defined by what any workforce member knew or should have known. The banking rule and the SEC rule start at a determination: that a notification incident occurred, or that an incident is material. A determination is a decision someone makes, and the date of that decision becomes the start of the clock.
Rules that start at a determination put weight on the decision process. If the process is slow or controlled by the wrong people, the clock starts late. In our reading, regulators reviewing a late notice will look at when the facts were known and when the determination was made. A recorded decision by a group that met promptly answers that question. A gap between knowledge in the security team and a determination elsewhere invites scrutiny.
Registers of open matters
The structural response is a register of open regulatory matters, kept by legal or compliance, with the topics each covers. Every significant incident should be checked against the register. If an incident touches a topic under inquiry, the lawyers handling that inquiry must be told. This is a simple control, and the 2016 events show what can happen without it.
Contracts and partners
Notice duties also come from contracts. A company that holds data for business customers may owe them notice under service agreements. A bank service provider or a health care business associate may have notice duties to the institutions it serves. These duties are easy to miss if the reportability decision considers only statutes and regulators. The register of obligations should include material contracts with notice clauses.
Regulators beyond the FTC
The same logic applies to every regulator. Consider a bank with an examination finding on information security, or a health system in an OCR investigation. A broker-dealer in an SEC or FINRA inquiry, or an agency in an inspector general audit, faces the same risk. An incident that touches the subject of an open matter may need to be disclosed in that matter, apart from any general notice rule.
The accountability: the company, an individual and the record
The fourth part concerns how responsibility was assigned. It took several forms over the years that followed, and each carries a lesson for governance.
New management and disclosure
Uber’s November 2017 statement came from new management. The non-prosecution agreement release says the breach was not reported to the FTC until new executive leadership was managing the company. It says the new leadership investigated and disclosed the breach to affected drivers, the public, law enforcement, foreign and domestic regulators, state attorneys general and the FTC. The agreement credited that conduct.
Disclosure here depended on a change of leadership. A governance structure that relies on new leaders to find old problems is fragile. Boards can ask, as a standing item, whether any significant incident in the past period was handled without the involvement of legal and compliance. The question is simple to ask and hard to answer falsely if a record exists.
Individual liability
The 2022 verdict assigned criminal liability to an individual. A federal jury convicted the former chief security officer of obstruction of FTC proceedings and misprision of felony. The Department of Justice release said the maximum sentences were five years and three years respectively. It is, in our reading, a clear signal that concealment of an incident from a regulator can be a personal criminal matter for the executive who conceals it.
The signal cuts two ways for security leaders. It warns against concealment. It also gives security leaders a reason to insist on shared decision-making. A leader who brings legal and compliance into every reportability decision is protected by the process. A leader who decides alone carries the decision alone.
The staff who knew
The trial evidence, as the Department of Justice described it, shows that the chief security officer was not the only person who knew. Employees working for him verified the theft. A subordinate drafted a summary. The intruders were identified and asked to sign new agreements. Knowledge was spread across a team, and it still did not reach the lawyers answering the FTC.
That pattern argues for a channel that lets staff raise a concern about incident handling outside their management line. The channel could run to the general counsel, the chief compliance officer or the audit committee. Staff need to know it exists and that using it is expected. A written statement to that effect, repeated in incident response training, costs nothing and changes what a concealment attempt has to overcome.
Corporate accountability
The company’s accountability came through agreements. The FTC’s expanded 2018 order added requirements, including that Uber notify the FTC of certain future incidents involving unauthorized access to consumer information. The non-prosecution agreement release says Uber also settled with the attorneys general of all 50 states and the District of Columbia for $148 million, with a corporate integrity program and breach response requirements.
Settlements as specifications
The settlements also describe a program in some detail. The FTC’s 2017 proposed order required a comprehensive privacy program and independent audits within 180 days and every two years after that for 20 years. The 2018 expansion added notice to the FTC of certain future incidents and retention of bounty records. The state settlement, as the Department of Justice release summarized it, required a corporate integrity program, specific data security safeguards, incident response and breach notification plans, and biennial assessments.
An organization with no order against it can read these terms as a list of what regulators asked for after a concealment. Incident response and notification plans, outside assessment, and records that a regulator can inspect appear in more than one of them. None is exotic. Each would have made the 2016 decision harder to keep inside one team.
The record as the defense
Across these outcomes, the record decided much. The trial turned on what the security team knew and when, and on what it told others. The company’s credit in the non-prosecution agreement turned on what new management did and disclosed. For any organization, the record of incident decisions is a governance asset. It should be complete, kept outside the security team, and reviewed.
What directors can ask
Directors do not need technical knowledge to test the structure. They can ask who decides whether an incident is reportable, and whether legal and compliance are always involved. They can ask to see the escalation triggers and a sample of recent decision records. They can ask whether any payment connected to an incident was made in the period, and who approved it. Each answer either shows a working process or exposes a gap.
Internal audit’s part
Internal audit can test the same structure on a cycle. It can sample incidents from the security team’s own tracking system and confirm that each one meeting a trigger reached the reportability group. It can compare bounty payments against incident records. It can check that the register of open matters is current. Testing from the security team’s records outward is the check that would catch an incident kept inside one function.
How the structure applies across sectors today
Notice rules have multiplied since 2016, as the hindsight section describes. Each has a clock that starts with a determination or with discovery, and each depends on how quickly the right people learn what happened. We take five kinds of organization in turn.
Banks and credit unions
Banks subject to the 2021 federal computer-security incident notification rule must notify their primary federal regulator within 36 hours of determining that a notification incident has occurred. The clock starts with a determination, which is exactly the classification step. Credit unions answer to their own regulator’s rules. In both cases, a reportability group that meets quickly keeps the clock from being missed or gamed.
Public companies and broker-dealers
Public companies, including listed banks and broker-dealer parents, face the SEC’s 2023 rule. An Item 1.05 Form 8-K is due four business days after the company determines an incident is material. Materiality is a disclosure committee judgment, not a security judgment. The Uber structure, in which security alone decided what an incident was, would put a public company at risk of a late or missing filing.
Health systems and their vendors
HIPAA covered entities must notify individuals without unreasonable delay and no later than 60 calendar days after discovering a breach of unsecured protected health information. The rule treats a breach as discovered on the first day it is known, or would have been known with reasonable diligence, to any workforce member or agent other than the person committing it. Knowledge inside the security team is knowledge of the covered entity, in our reading.
That rule closes the gap the Uber case exposed, at least on paper. A health system cannot argue that the clock did not start because only its security staff knew. The practical consequence is that security staff knowledge must reach privacy and legal quickly, or the 60-day clock runs without anyone responsible for notice knowing it has started.
Service providers to regulated institutions
The 2021 banking rule also reached bank service providers, which must notify affected banking organization customers of certain incidents as soon as possible. A technology vendor serving banks therefore needs its own reportability process, because its determination starts its customers’ clocks. The same is true for vendors to health systems and agencies under their contracts. A vendor that lets security alone classify incidents passes the Uber structure on to every customer.
Public agencies
Federal agencies report incidents under federal policy, and state and local agencies under state law and their own policies. Agencies also answer to inspectors general and legislatures. The same structural risk applies: an incident that touches the subject of an audit or legislative inquiry may need to be disclosed there. A register of open oversight matters serves an agency the way a register of regulatory matters serves a bank.
Objections, in order
The argument that the reporting decision needs owners outside security faces objections. We take the strongest six in order.
1. The problem was individual misconduct, not structure
The jury found individual criminal conduct. We do not dispute it. But structure determines how far one person’s choices can reach. We think a structure in which legal and compliance must classify every extortion demand would have made the 2016 event much harder to keep inside one team. Good structure limits the damage individual misconduct can do.
2. Lawyers will slow incident response
The reportability group need not touch containment or investigation. It decides one question on the facts security provides. If it meets within hours, it costs little time. The alternative, as the Uber case shows, can cost years of proceedings.
3. Bug bounty programs will suffer if every payment needs approval
Routine bounty payments for flaw reports need not change. The rule we suggest applies to payments connected to data access or extortion, which should be exceptional in a well-run program. Separating those from routine research protects the program’s legitimacy.
4. Notice rules now force disclosure, so the structure matters less
Notice rules start their clocks with a determination or discovery. The structure decides when that happens and whether it is recorded honestly. Stronger rules raise the stakes of the classification step. They do not remove it.
5. Small organizations cannot staff such a group
A small credit union, clinic or county may not have in-house counsel. It can still name an outside counsel, a compliance lead and an executive who must be called for defined triggers. The group can be three people and a phone list. What matters is that the decision does not rest with one function.
6. Disclosure invites punishment, so caution about reporting is rational
Disclosure does carry costs. The Uber record shows the costs of concealment were larger and fell on both the company and an individual. The non-prosecution agreement credited new management’s prompt disclosure and cooperation. In our reading, the record points one way: early, documented disclosure decisions are the safer course.
What a compliance officer could check
The Uber case suggests a short set of checks for any regulated organization:
- Name the group that decides whether an incident is reportable, with legal and compliance as members.
- Write escalation triggers that bypass security’s discretion, including any extortion demand and any confirmed access to personal data.
- Require legal and finance approval for any payment connected to an incident, including bounty awards tied to data access.
- Keep a register of open regulatory, audit and oversight matters, and check each significant incident against it.
- Record each reportability decision, with the facts, the analysis and the people involved, outside the security team’s own systems.
A tabletop exercise is a cheap way to test the checks. The scenario can follow the 2016 pattern: an outsider emails a security leader, claims to hold customer data and asks for payment. The exercise should test whether the demand reaches legal and compliance the same day, whether anyone checks the register of open matters, and whether a proposed payment is stopped for approval. Each failure in the exercise is a gap to fix before a real demand arrives.
The exercise should include the people who answer regulators, as well as the incident response team. Their part is to say whether the incident touches an open matter and what the organization has already told the regulator. If they are absent from the exercise, they are likely absent from the real event too.
These checks are inexpensive. Each addresses a failure point the 2016 events exposed. Together they move the reporting decision from one function to a group that can be held to account, which is where it belongs.
With hindsight
The FTC moved first. On April 12, 2018, it announced that Uber had agreed to expand the proposed 2017 settlement. The Commission said it had learned, after announcing the original settlement, that Uber had failed to disclose a significant breach in 2016, in the midst of the FTC’s investigation. The revised complaint alleged that intruders used an access key an Uber engineer had posted on a code-sharing website to download unencrypted files from cloud storage.
The revised complaint gave the FTC’s count of what was taken: more than 25 million names and email addresses, 22 million names and mobile phone numbers, and 600,000 names and driver’s license numbers of U.S. drivers and riders. It alleged the $100,000 bug bounty payment and the failure to disclose until November 2017. The expanded order could subject Uber to civil penalties if it failed to notify the FTC of certain future incidents, and required bug bounty record retention.
The Department of Justice resolved the company’s criminal exposure on July 22, 2022. Uber entered a non-prosecution agreement and admitted that its personnel failed to report the November 2016 breach to the FTC despite the pending investigation. The release said the intruders had used stolen credentials to reach a private source code repository and obtain the access key. It credited new leadership’s disclosure, investment in compliance and security, the October 2018 FTC order, cooperation, and the $148 million state settlement.
The individual case followed. On October 5, 2022, a federal jury convicted the former chief security officer of obstruction of FTC proceedings and misprision of felony. The Department of Justice release described the evidence, including the non-disclosure agreements and the payment in bitcoin. It also said the two intruders had pleaded guilty in 2019 to computer fraud conspiracy, and that after the Uber matter they had attacked and tried to ransom another company.
Notice rules tightened across sectors later. On November 18, 2021, the federal banking agencies approved a rule requiring banking organizations to notify their primary federal regulator no later than 36 hours after determining that a significant computer-security incident had occurred. The rule also required bank service providers to notify affected bank customers of certain incidents. Compliance was required by May 1, 2022.
On July 26, 2023, the SEC adopted rules requiring public companies to disclose material cybersecurity incidents on a new Item 1.05 of Form 8-K. The filing is generally due four business days after the company determines an incident is material. The rules also added Regulation S-K Item 106, requiring annual disclosure of cybersecurity risk management and of board oversight. Both rules start their clocks with a determination, the step the Uber case showed can be controlled by the wrong people.
HIPAA’s breach notification rule treats knowledge by any workforce member, other than the person committing the breach, as discovery by the covered entity. Its current text, which we checked, sets that standard and the 60-day outer limit for notice to individuals. In health care, the Uber structure would not have stopped the clock, in our reading. It would have let the clock run unseen.
What held up is the structural reading. Each later rule assumes that an organization will determine, promptly and honestly, what an incident is. The Uber case showed what happens when that determination is made by the people whose own conduct it judges. Organizations that in 2026 still let security alone classify incidents are relying on the integrity of individuals where a process could do the work.
Spotted an error? Tell us and we will check it and correct the article where needed.