Analytical report · Hate speech · 2022–2025

Public Discourse in Ukraine Concerning LGBTIQ+ People, 2022–2025: Hate Speech-Related Risks and Architecture of Response

This analytical report by Bureau "We Are!" examines recurring patterns of public discourse concerning LGBTIQ+ people in Ukraine, hate speech-related risks and the minimum institutional architecture required for response. It does not propose a censorship model; it sets out an evidence-based, proportionate approach compatible with Article 10 ECHR.

Key facts

Analysis period
2022–2025

With contextual calibration of the 2016–2021 institutional layer.

Media corpus
N=35

A coded set of open materials from one online outlet used as a reproducible thematic case study.

Institutional subsample
107 records

Decisions and appeals of local councils from 2016–2019 across 21 regions of Ukraine.

Method
3 evidence streams

Content, harm and risk, and response process.

Key principle
Corpus ≠ prevalence

The report does not measure overall prevalence and does not legally qualify content.

Scope of conclusions

The report describes recurring discourse patterns and response parameters. It does not measure overall prevalence, replace the legal qualification of competent authorities or propose prior control of expression.

  • Not about censorship.
  • Not about automatic punishment.
  • About evidence-based and proportionate response.

Why Ukraine needs an architecture for responding to hate speech-related risks

The public-policy problem arises not only when harmful discursive patterns recur, but when institutions lack a predictable procedure for responding to them. In wartime, this affects the social acceptability of discrimination, barriers to justice, community safety and trust in the state.

The report’s central conclusion is that risk assessments alone are not enough. Without aligned roles, thresholds and procedures, they do not become governable practice.

Three evidence streams

The methodology combines three linked strands that separate content description from risk assessment and the response process.

1

Content stream

Records documented message content and recurring interpretive patterns without attributing hidden intent.

2

Harm and risk stream

Assesses potential mechanisms of social exclusion or increased harm risk using an operational severity scale from 0 to 4.

3

Response stream

Tracks verified indicators of institutional or platform response: response_status, responder, response_action, first_response_date and lag_days.

Content stream
Harm & risk stream
Response stream
Architecture of response

How risk configurations changed in 2022–2025

2016–2019

Institutional calibration layer

Local council decisions and appeals, central-government documents and a court case are used as indicators of roles, competence boundaries and procedural thresholds, not as evidence of prevalence.

2022

Securitisation of equality

In wartime, equality is more often interpreted through security arguments, loyalty and “social usefulness”. The key gap is fragmented response data.

2023

Institutional articulation without aligned response

Standards of responsible public communication become more visible, but normative indicators are not backed by transparent thresholds or structured response data.

2024

High visibility and incomplete alignment

Several institutional response indicators emerge, including in the regulatory sphere, but a reproducible cross-sector model is still missing.

2025

A test of institutional inertia

Recurring patterns of social exclusion remain visible, while the central challenge becomes the shift from isolated instruments to an aligned operational architecture of response.

What the coded corpus shows

The charts below show distributions only within the coded media corpus. They must not be interpreted as estimates of overall prevalence in Ukrainian public discourse or as representative statistics.

Narrative codes by year

2022
N6 3N1 2N3 2111
2023
N6 3N3 211
2024
N7 611
2025
N7 6N5 211
N1N2N3N4N5N6N7N8N9N10
In 2024–2025, the N7 line became more visible within the corpus, linked to narratives about “child protection” and sexual danger. This is not a measure of prevalence in society; it is only a distribution within the constructed corpus.
Show data as a table
Narrative codes by year
YearN1N2N3N4N5N6N7N8N9N10
20222121130000
20231020130000
20240000006110
20250000216001

HS codes by year

2022
HS3 5HS4 5HS2 4HS7 41
2023
HS4 4HS7 4HS2 3HS3 31
2024
HS2 8HS3 21
2025
HS2 7HS7 4HS4 3HS3 2
In 2024–2025, the share of HS2 increased within the corpus, reflecting criminalising or danger-based framing. Because multiple coding is used, the sum of codes may exceed the number of materials in a given year.
Show data as a table
HS codes by year
YearHS1HS2HS3HS4HS7
202214554
202313344
202408210
202507234

D codes by year

2022
D4 5D2 4D3 3
2023
D4 4D2 3D3 3
2024
D2 7D4 51
2025
D2 8D4 51
D2, associated with contextual manipulation, becomes more visible in the 2024–2025 corpus, while D4 remains consistently present.
Show data as a table
D codes by year
YearD2D3D4
2022435
2023334
2024715
2025815

Severity distribution

2022
1:32:53:2
2023
1:22:41
2024
12:43:3
2025
1:22:8
Severity level 2 is the most frequent in the corpus, especially in 2025. The operational severity scale is not a scale of legal consequences and does not create automatic grounds for intervention.
Show data as a table
Severity distribution
YearLevel 1Level 2Level 3Level 4
20223520
20232410
20241430
20252800

Threshold matrix: relating risk to response

The report’s MVP model proposes a threshold-based approach: the higher the operational level of harm risk, the stricter the requirements for verifying grounds, procedural safeguards and responder competence. The matrix does not create automatic legal consequences and does not replace decisions by competent authorities.

0–1Monitoring and counter-communication
2Administrative and regulatory signals
3Formalised threshold analysis
4Most intensive interventions only as ultima ratio
LevelThresholdType of responseActorsMinimum metrics
0–1 · Low riskIsolated cases that do not approach the high threshold of incitement.Monitoring, contextual explanation, ethical self-regulation mechanisms and counter-communication.CSOs, media, platforms and self-regulatory mechanisms.response_status, responder.
2 · Medium riskSystematic practices that may contribute to an exclusionary environment but do not involve a direct call to violence.Administrative and regulatory signals, orders and media regulation based on transparent criteria.Regulatory bodies and executive authorities.response_status, responder, response_action.
3 · Elevated riskHigh likelihood of harm; requires deeper analysis of speaker status, reach, form and context.Formalised inter-institutional threshold analysis within existing mandates.Regulatory bodies and other competent institutions.response_status, responder, response_action, first_response_date.
4 · Highest riskThe gravest cases, which may approach public incitement to violence or discrimination with a high risk of real harm.Intensive interventions, including criminal-law mechanisms, only as ultima ratio.The judiciary and competent response actors.response_status, responder, response_action, first_response_date, lag_days.

What the report proposes

The recommendations are aimed not at mechanically expanding restrictive tools, but at improving the quality, predictability and verifiability of response.

Core system

Minimum data standard

Record response_status, responder, response_action, first_response_date and lag_days for cases that enter the response process.

Transparent thresholds

Publish guidance for matching harm-risk levels with types of response instruments.

Functional allocation of roles

Align which tasks belong to regulatory, administrative, human-rights, self-regulatory and judicial mechanisms.

Conditions for resilience

Regular analytics

Update data monthly and review thresholds, the codebook and procedures quarterly.

Media and platform self-regulation

Support non-punitive tools for responsible communication, moderation and counter-communication.

Training response actors

Integrate the harm & risk stream, threshold approach and Article 10 ECHR standards into training for institutions and partners.

Who the report is for

Public institutions

For calibrating roles, thresholds, routing and procedural response metrics.

International partners

For assessing implementation gaps, democratic resilience and support needs for a human rights-based response architecture.

Media and platforms

For understanding the boundaries of responsible public communication, self-regulation and proportionate moderation.

Human rights organisations

For documenting risks without reproducing harmful content or substituting legal qualification.

Researchers and analysts

For using a reproducible coding scheme, neutral excerpt fields and procedural indicators.

Download the report and access related tools

Full PDF report

Full text of the analytical report.

Download PDF

Report presentation

Visual summary of the key findings and response architecture.

Open presentation

Presentation No. 13597

Separate presentation on draft law No. 13597 and response standards.

Open presentation

Report an incident

Submit information through the secure monitoring form.

Open form

Bureau analytics

Explore other analytical materials.

All materials

Partnership

Contact us about cooperation or support.

Contact us

Questions and answers

Does the report measure the prevalence of hate speech in Ukraine?

No. The report analyses a reproducible corpus of open materials and procedural response indicators. It cannot be used as a statistical estimate of the overall prevalence of the phenomenon in Ukrainian society.

Does the report provide a legal qualification of specific content?

No. The operational HS, D and severity codes are used for analytical risk classification. They do not replace a decision by a court, regulator or other competent authority.

Does the report propose censorship?

No. The report is explicitly grounded in Article 10 ECHR and the principles of legality, legitimate aim, necessity and proportionality. The most intensive forms of intervention are considered only as ultima ratio.

Why does the page not reproduce examples of hate speech?

To avoid recirculating harmful content. The report methodology uses neutral summaries without reproducing stigmatizing wording and without collecting personal data beyond what is necessary.

What is an MVP architecture of response?

It is a minimum viable model for recording, routing and analysing response processes: who responds, with which instrument, when and under what procedural logic.

How can this report be used?

It can be used for advocacy, training, policy development, donor applications, institutional dialogue, media self-regulation and the design of monitoring systems.

Build a response architecture that institutions can trust

Bureau "We Are!" is open to cooperation with institutions, donors, media and human rights organisations working with evidence-based, proportionate and human rights-based response models.