A data breach has occurred. Do you need to notify the regulator, the people affected, both, or neither? And can you show how you reached that decision?
This update is for businesses that handle personal data or operate information systems in the Baltics and elsewhere in the European Union (EU).
Quick Navigation
- Lithuania: New guidance on assessing data breach risks
- Latvia: New data center rules and deadlines for class A systems
- Digital Omnibus: Council position on incident reporting and cookies
- Estonia: Security audits for essential entities and voluntary near-miss reporting
- AI Act: Reference to quality management standard published
- Smart glasses: Data protection and workplace risks
- Estonia: Draft guidance on publishing payment default data
- Our view
- How ECOVIS ProventusLaw can help
- Why ECOVIS ProventusLaw
Lithuania: New guidance on assessing data breach risks
Date: October 26, 2026 | Source: Valstybinė duomenų apsaugos inspekcija (State Data Protection Inspectorate, VDAI)
Link (recommendation) | Link (VDAI announcement)
On October 26, 2026, the Valstybinė duomenų apsaugos inspekcija (State Data Protection Inspectorate, VDAI) published a recommendation, version 1 dated October 23, 2026, on how to assess the risk posed by a personal data breach. The assessment decides whether a breach must be notified to VDAI under Article 33 of the General Data Protection Regulation (GDPR) and to the persons concerned under Article 34 GDPR. The controller must notify VDAI without undue delay and, where feasible, within 72 hours of becoming aware of the breach, unless the breach is unlikely to result in a risk to individuals’ rights and freedoms. If notification is made after 72 hours, it must include the reasons for the delay. The persons concerned must be informed where the breach is likely to result in a high risk, unless an exception in Article 34(3) applies.
The recommendation sets out:
- Factors to assess: the type of breach and whether it affects confidentiality, integrity or availability (a breach of availability can include temporary unavailability after a cyber incident), and the nature, sensitivity and volume of the data. Controllers should also consider how easily individuals can be identified, the severity and duration of the consequences, and whether vulnerable groups are affected. Encryption reduces risk only if a strong algorithm is used and the keys have not been compromised.
- A risk matrix combining severity (low, medium, high) and likelihood (low, medium, high), with three outcomes: no risk (document only), risk (document and notify VDAI) and high risk (also inform the persons concerned, unless an Article 34(3) exception applies). The final level remains the controller’s decision.
- Five worked examples, from a database error that gave 53 contracts the same number (no risk) to a compromised employee mailbox (notify VDAI) and the mass copying of 58,000 records (inform the persons concerned).
- Reassessment and AI: controllers should reassess when new facts appear, analyze immediate, medium-term and long-term scenarios for complex breaches and take into account that AI makes leaked data easier to combine and misuse.
Why it matters
VDAI expects the assessment to be based on facts and technical evidence, such as logs showing whether data were actually copied, and to be documented. A decision not to notify needs the same record as a decision to notify, and a later change in the facts can turn a documented breach into one that must be notified to the persons concerned.
Recommended actions
Data controllers should:
- align the internal breach procedure and risk assessment template with VDAI’s factors and matrix;
- make sure the incident team can obtain technical evidence (logs, access records) quickly enough to meet the 72-hour deadline;
- add a reassessment step for breaches where new facts may raise the risk level;
- train the people responsible for logging and assessing breaches on the recommendation and its examples.
Latvia: New data center rules and deadlines for class A systems
Date: October 1, 2026 | Source: Cabinet of Ministers Regulation No 602 of September 29, 2026 on information system placement and data center security requirements (Latvia)
Link
The Latvian Cabinet of Ministers adopted Regulation No 602 on September 29, 2026 under the National Cybersecurity Law, which transposes Directive (EU) 2022/2555 on measures for a high common level of cybersecurity across the Union (NIS2 Directive). It was published and entered into force on October 1, 2026. It applies to essential and important entities and to owners and lawful possessors of information and communications technology (ICT) critical infrastructure, insofar as Implementing Regulation (EU) 2024/2690 does not already regulate the matter for digital infrastructure and ICT service providers.
Key rules include:
- Location. Unless the regulation provides otherwise, information systems must be hosted in data centers in a NATO, EU or European Free Trade Association (EFTA) country or one of NATO’s four Indo-Pacific partners, and data must be processed there. Stricter placement rules apply to owners and lawful possessors of critical ICT infrastructure: depending on the arrangement, they must notify the Constitution Protection Bureau before the placement or obtain its positive opinion. Using a cloud service does not exempt the entity from these rules.
- Class A systems. The most critical systems (class A under the minimum cybersecurity requirements) may be hosted outside the entity’s own infrastructure only in a data center whose conformity is confirmed by an auditor’s opinion, renewed at least every two years, or by a recognized certificate such as Uptime Institute Tier III or IV, TIA-942 Rated-3 or 4, EN 50600 or ISO/IEC 22237 at the required classes. An Uptime Institute certificate must be accompanied by an additional auditor’s opinion on the organizational, technical and physical security checks in Annex 1, valid for two years. An entity that is not an owner or lawful possessor of critical ICT infrastructure may use a data center without confirmed conformity by way of exception, if it documents the need, assesses the risks, sets risk mitigation measures and obtains prior approval from the competent supervisory authority. Separate exceptions apply to certain temporary placements and backup copies. For a class A system hosted outside Latvia, the contract must give the Latvian authorities audit, supervision and inspection rights, to the extent technically feasible and permitted by the law of the country concerned.
- Contracts and reporting. Outsourcing contracts must oblige the data center operator to meet the regulation and confirm the location. The operator must inform the entity within five working days of any issue affecting compliance, and the entity reports its data centers to the supervisory authority in its self-assessment report.
- Telemetry. Entities with class A systems must give the competent incident response institution access to telemetry data at source and in real time, within one minute of an event.
- Audit benchmarks. Annex 1 sets out what the data center auditor checks. This includes annual availability of at least 99.982%, incident response starting within 15 minutes of detection, generator fuel for at least 24 hours, physical access logs kept for at least 18 months, staff security checks at least every two years, and written notice to clients at least five working days before maintenance that may affect their equipment. For the data center’s management information systems, the checklist sets a recovery time objective of up to one hour and a recovery point objective of up to 15 minutes.
- Transition. No new non-compliant class A placement may start after October 1, 2026, except under the exceptions for approved placements, temporary placements and backup copies. Existing class A placements must be notified by December 31, 2026 to the data center operator, unless the entity is itself the operator, and, for owners and lawful possessors of critical ICT infrastructure, also to the Constitution Protection Bureau. Non-compliant placements may continue until December 31, 2027 if the risks are managed and a compliance plan is prepared by December 31, 2026. Contracts may not be extended to keep a non-compliant placement after that date.
Why it matters
Covered entities need to review where their information systems are hosted, where their data are processed, and whether their data centers meet the applicable security requirements. The location rule applies to all information systems from October 1, 2026 and the regulation gives no transitional period for it, except for class A systems: existing class A placements that do not meet the requirements may continue only until December 31, 2027, under a compliance plan.
Recommended actions
Entities covered by the Latvian National Cybersecurity Law should:
- list their information systems with their class, hosting provider, data center and processing location;
- notify existing class A placements to the data center operator by December 31, 2026 and, where a placement does not meet the requirements, prepare a compliance plan by the same date;
- obtain the auditor’s opinion or certificate for the data centers used for class A systems, including the additional auditor’s opinion required with an Uptime Institute certificate;
- update hosting and cloud contracts with location, audit rights, five-day notice and telemetry access terms.
Digital Omnibus: Council position on incident reporting and cookies
Date: October 21, 2026 | Source: Council of the European Union – draft legislation
Link
The Council’s mandate for negotiations with the European Parliament on the Digital Omnibus proposal (2025/0360(COD)) was agreed by the Permanent Representatives Committee on October 21, 2026 and published on October 22, 2026 as Council document 14756/26. Among other acts, the proposal amends the GDPR, the ePrivacy Directive, the NIS2 Directive, the Digital Operational Resilience Act (DORA), the Critical Entities Resilience Directive and the Data Act, and folds the Data Governance Act and the Open Data Directive into the Data Act. The text is not yet law: it is the Council’s position for negotiations with the European Parliament, and the final rules may differ.
Key points of the Council text:
- Breach notification. Controllers would notify a supervisory authority only of breaches likely to result in a high risk, where feasible within 96 hours, with the support of a national entry point set up under the NIS2 Directive. Until that point exists, notifications go directly to the authority. The European Data Protection Board (EDPB) would publish lists of circumstances that are and are not likely to result in a high risk.
- Cookies. Unlike the Commission, which proposed moving the rules into the GDPR, the Council keeps them in Article 5(3) of the ePrivacy Directive. Consent remains the rule, with exemptions for transmission, services explicitly requested by the user, anonymous aggregated audience measurement, audience measurement under Article 24 of the European Media Freedom Act, security and measuring contextual advertising. Users must be able to refuse with a single click or equivalent means, and after a refusal the provider may not request consent again for the same purpose for at least four months.
- DORA incident reporting. Financial entities would report major ICT-related incidents to their competent authority with the support of the same national entry point. The European Union Agency for Cybersecurity (ENISA) would run an information point that directs entities to the right reporting channel but does not receive reports.
- AI. Processing of personal data for the development and operation of AI systems and models could rely on legitimate interests, subject to safeguards and the right to object.
Why it matters
If adopted in this form, the changes would reduce the number of breach notifications but not the duty to assess and document every breach. Cookie banners would have to offer a single-click refusal, and after a refusal providers could not ask again for consent for the same purpose for four months.
Recommended actions
Data controllers and financial entities should, as preparatory steps only:
- keep notifying breaches under the current 72-hour rule until the amendments are adopted and apply;
- review cookie banners against the single-click refusal and the exemptions in the Council text;
- track the trilogue outcome on the reporting channel for GDPR breaches and DORA incidents.
Estonia: Security audits for essential entities and voluntary near-miss reporting
Date: October 6, 2026 | Source: Act amending the Cybersecurity Act, RT I, 06.10.2026, 1 (Estonia)
Link (amending act) | Link (consolidated Cybersecurity Act)
The Riigikogu adopted an act amending the Estonian Cybersecurity Act on September 23, 2026. It was published in Riigi Teataja on October 6, 2026 and entered into force on October 16, 2026. It completes the transposition of the NIS2 Directive’s supervision rules.
Key changes include:
- Near-miss incidents. The act defines a near-miss cyber incident as an event that could have compromised the availability, authenticity, integrity or confidentiality of data or services but was prevented or did not materialize. Service providers may voluntarily notify the Information System Authority (RIA) of near misses, incidents, vulnerabilities and threats, and the incident register now covers near misses.
- Supervision of essential entities. RIA may carry out on-site and remote supervision, including spot checks of essential entities, and order ad hoc security audits, for example after a significant incident or a breach of the NIS2 implementing acts.
- Regular audits. RIA may carry out regular security audits of essential entities once every three years, unless an entity is already required to be audited regularly under the cybersecurity requirements set under section 7(5) of the act.
- Definitions. The definitions of cyber incident and security audit are updated.
Why it matters
Essential entities may face regular RIA audits every three years, unless they are already audited under section 7(5), and should keep evidence of their security measures ready. Recording near misses can also help entities assess risks, even though reporting them is voluntary.
Recommended actions
Essential and important entities in Estonia should:
- check whether an existing audit cycle under section 7(5) covers the new three-year audit;
- prepare audit evidence for RIA spot checks and audits;
- add near misses to incident logging and decide when to report them voluntarily.
AI Act: Reference to quality management standard published
Date: October 14, 2026 | Source: Commission Implementing Decision (EU) 2026/2274 of October 12, 2026
Link
The European Commission adopted Implementing Decision (EU) 2026/2274 on October 12, 2026. Published in the Official Journal and in force on October 14, 2026, it publishes the reference of EN 18286:2026 “Artificial intelligence – Quality management system for EU AI Act regulatory purposes”, developed under the Commission’s standardization request of June 23, 2025.
Under Article 40(1) of the AI Act, high-risk AI systems or general-purpose AI models that conform with harmonized standards published in the Official Journal are presumed to conform with the requirements in Section 2 of Chapter III (or, for general-purpose AI models, the obligations in Chapter V), to the extent the standards cover them. According to the decision, EN 18286 responds to the request for a standard on quality management systems for providers of AI systems, including post-market monitoring, and the presumption applies from the date of publication of its reference.
Article 17, in Section 3 of Chapter III, sets out the quality management system duty. It covers, among other things, a regulatory compliance strategy, design and quality control, testing, data management, the risk management system, post-market monitoring, serious incident reporting, record-keeping and an accountability framework. The system must be applied proportionately to the provider’s size. Under Article 17(4), a provider that is a financial institution subject to EU internal governance rules meets the duty, except for risk management, post-market monitoring and incident reporting, by complying with those rules, taking harmonized standards into account.
Why it matters
The decision does not state which AI Act provisions EN 18286 covers. Because Article 40(1) refers to the requirements in Section 2 while the quality management system duty is in Article 17, providers should check the requirements mapped in the standard itself before relying on the presumption. The Commission describes the standard as designed for the Article 17 quality management requirements.
Recommended actions
Providers of high-risk AI systems should:
- compare their existing quality management system with EN 18286:2026 and document any gaps;
- use conformity with the standard in technical documentation and the conformity assessment.
Businesses that buy high-risk AI systems should, as deployers:
- ask providers whether their quality management system follows EN 18286.
Smart glasses: Data protection and workplace risks
Date: October 13, 2026 | Source: European Data Protection Board (EDPB), Statement 1/2026
Link
The EDPB adopted Statement 1/2026 on October 13, 2026, setting out preliminary reflections on smart glasses, meaning glasses with cameras, microphones and AI features that process personal data of users and bystanders.
Key points include:
- Controllers. The user is a controller when using the glasses for professional, official or political purposes or sharing recordings publicly, since the household exemption is narrow. Producers and providers of linked apps or AI services may be independent or joint controllers when they determine the purposes and means of processing. Where they are controllers, they must meet their own obligations, such as transparency, and cannot rely on users to do so.
- Workplace use. Use at work raises concerns about employee monitoring and confidentiality, including trade secrets, and may be inappropriate or prohibited under national law or local rules.
- Transparency. Privacy notices alone do not reach bystanders. In-the-moment measures are expected by default, such as visible LED indicators that block recording if covered, audio signals and design cues, which must be effective and hard to circumvent.
- Legal basis and minimization. Consent from bystanders is difficult to obtain, and consent obtained after recording cannot provide the legal basis for the recording that has already taken place. Legitimate interest needs careful balancing, especially for commercial further use or AI training. The EDPB recommends limiting continuous recording, live streaming and retention, blurring bystanders and processing on the device. For glasses designed primarily to provide real-time assistance, persistent storage of captured audio, images or video should be disabled by default unless it is required for the intended functionality or deliberately activated by the user, subject to appropriate safeguards.
Why it matters
Employers need to consider these risks whether they provide smart glasses or employees bring their own. Recording colleagues, customers or confidential information at work can breach the GDPR and confidentiality duties at the same time.
Recommended actions
Data controllers, in particular employers, should:
- decide whether smart glasses are allowed at work and set the rules in internal policies;
- carry out a data protection impact assessment (DPIA) before any business use of smart glasses;
- check the privacy features of devices before procurement.
Estonia: Draft guidance on publishing payment default data
Date: October 14, 2026 | Source: Andmekaitse Inspektsioon (Estonian Data Protection Inspectorate, AKI) – draft guidance
Link (draft guide) | Link (AKI announcement)
The Andmekaitse Inspektsioon (Estonian Data Protection Inspectorate, AKI) opened a public consultation on its draft guide to publishing payment default data on October 14, 2026. The consultation closed on November 15, 2026, and AKI plans to publish the final guide after reviewing the feedback. The guide is a draft and may change.
Key points of the draft include:
- Roles and legal basis. Creditors that transfer data to a private payment default register and the register keeper are separate controllers. A private register relies on legitimate interests, while creditors in the financial sector assessing creditworthiness may rely on Article 6(1)(e) GDPR together with section 10 of the Estonian Personal Data Protection Act.
- Purpose. Publication may not be used to pressure debtors to pay, and threatening to publish a time-barred claim is contrary to good morals.
- Accuracy and notice. The register and the creditor share responsibility for accuracy. Where there is doubt, the register must check the underlying documents; an invoice alone is not enough. The creditor must notify the debtor before transferring the data, and a general clause in a loan agreement signed long ago is not enough.
- Retention. The five-year period in section 10 of the Personal Data Protection Act is a guide, not an automatic publication period. The draft considers the full period appropriate only for the most serious defaults, and publication beyond five years on the basis of legitimate interests requires particularly compelling reasons. Time-barred claims may generally not be published, and data must be removed once an insolvency discharge is no longer published.
- Recipients and employers. Registers must check each recipient’s purpose and legal basis. Employers may consider checking payment defaults only for positions involving significant financial responsibility or where the check has a direct and justified connection to the duties of the role. They also need an appropriate legal basis, and the processing must be necessary and proportionate.
Why it matters
The draft would tighten practice for lenders, telecom and utility companies, debt collectors and register keepers in Estonia, as well as employers that use payment default checks in recruitment.
Recommended actions
Creditors, register keepers and recipients of payment default data in Estonia should:
- review debtor notices, accuracy checks and retention periods against the draft;
- limit payment default checks in recruitment to roles with significant financial responsibility or a direct and justified connection to the duties of the role.
Our view
For Lithuanian controllers, the VDAI recommendation is the development of the month. Controllers should test their breach procedures against VDAI’s five examples and update them if they would reach a different result. Those that rarely notify should also check whether this reflects few incidents or an assessment threshold set too high.
In Latvia, the data center regulation is the deadline to plan for. Hosting contracts that run beyond 2027 should be reviewed now.
The Digital Omnibus may raise the notification threshold and extend the deadline, but the Council text is a negotiating mandate, not law. A breach assessment built on VDAI’s method will remain useful whatever the final rules.
How ECOVIS ProventusLaw can help
We advise data controllers and ICT-dependent businesses on data protection and operational resilience.
- Breach management: breach procedures, risk assessment templates and support with notifications to VDAI and data subjects.
- GDPR policies and procedures: drafting and updating privacy notices, records and internal procedures.
- Cookies and online tracking: reviewing consent banners and tracking technologies.
- DPIA: data protection impact assessments for new products and processing.
- Data protection officer: external data protection officer services, including advice on GDPR obligations, cooperation with supervisory authorities and staff training.
- DORA: ICT risk management, incident reporting, ICT third-party risk and resilience testing.
- NIS2 and cybersecurity: scoping under the Lithuanian, Latvian and Estonian cybersecurity laws, hosting and data center requirements and preparing for supervisory audits.
- AI Act: quality management systems, conformity documentation and AI procurement terms.
Contact our data protection team to align your breach procedure with VDAI’s new recommendation.
Why ECOVIS ProventusLaw
Our technology team advises on data protection and DORA requirements, including ICT risk management, incident reporting, ICT third-party risk and digital operational resilience testing. This is our advantage: one team covers both data protection and operational resilience, so your procedures and reports stay consistent. This matters now that GDPR breaches and DORA incidents may come to share a reporting channel.
The firm is ranked by IFLR1000 for M&A, Financial services regulatory, and Banking and finance. The Legal 500 ranks the firm for TMT, Tax, Real estate and construction, Fintech, Employment, Dispute resolution, and Banking, finance and capital markets. The firm is also ranked by Chambers and Partners in its FinTech and Europe guides.
Data protection and ICT are a core part of our fintech practice. Credit institutions, payment and e-money institutions, and other companies in this sector deal with GDPR and DORA every day. Read more in our insight on data protection in fintech companies. We also work with telecommunications, defense technology and logistics businesses.
Our advice is practical. We prepare procedures your teams can use, documents that match how your systems work, and evidence you can show a supervisory authority.
ECOVIS ProventusLaw operates across Lithuania, Latvia and Estonia and is part of the ECOVIS International network, present in more than 90 countries worldwide. This allows us to support clients on cross-border processing and multi-jurisdictional matters.
LT
RU
CN
DE