Threat Assessment Methodology
This document explains how threats in the Threat Assessment are identified, evaluated, and assigned probability ratings. Transparency about methodology enables scrutiny and challenge.
1. Assessment Framework
1.1 Threat Categories
Threats are categorised across three dimensions:
| Dimension | Description | Evidence Types |
|---|---|---|
| Capability | Technical ability to execute the threat | Architecture documentation, technical analysis, demonstrated incidents |
| Legal Authority | Lawful basis to compel action | Legislation text, court rulings, DOJ guidance, transparency reports |
| Intent | Willingness to use capability | Historical precedents, official statements, policy patterns |
Assessment Principle: A threat requires all three elements. Capability without legal authority is constrained. Legal authority without capability is theoretical. Capability and authority without intent is latent but may activate under changed circumstances.
1.2 Probability Rating Framework
Probability ratings in the Threat Assessment use the following definitions:
| Rating | Definition | Indicative 10-Year Probability |
|---|---|---|
| HIGH | Strong evidence of intent; capability and authority confirmed; historical precedent exists | >50% |
| MEDIUM | Capability and authority confirmed; intent uncertain but plausible | 20-50% |
| LOW | Capability exists; authority may be contested; intent unclear | 5-20% |
| MINIMAL | Theoretical capability; significant barriers to execution | <5% |
2. Data Sources and Evidence Base
2.1 Primary Sources
| Source Type | Examples | Use in Assessment |
|---|---|---|
| Legislation & Regulation | CLOUD Act (H.R.4943), IEEPA, FISA Section 702 | Legal authority confirmation |
| Court Rulings | Schrems II (C-311/18), Microsoft Ireland case | Legal interpretation, precedent |
| Government Agency Guidance | DOJ CLOUD Act FAQ, Cross-Border Data Forum FAQs | Implementation interpretation |
| Provider Documentation | AWS CLOUD Act page, AWS Global Services | Technical architecture, capability |
| Transparency Reports | Microsoft Government Requests, Google Transparency Report | Request volumes, compliance rates |
2.2 Secondary Sources
| Source Type | Examples | Use in Assessment |
|---|---|---|
| Academic Analysis | Lawfare, CSIS, university law reviews | Legal interpretation, policy analysis |
| Congressional Research | CRS Reports on CLOUD Act, FISA | Authoritative legislative analysis |
| Market Research | Canalys, Synergy Research, Gartner | Market share data |
| News & Investigative Journalism | Reuters, Financial Times, The Register, Ars Technica | Incident reporting, policy announcements |
3. Evidence Basis by Threat
For each threat scenario in the Threat Assessment, the following evidence is cited:
3.1 Scenario A: Trade Negotiation Leverage
Evidence for HIGH Probability Rating
| Capability | Confirmed - Provider architecture enables service termination |
|---|---|
| Legal Authority | Confirmed - IEEPA grants broad presidential powers |
| Intent Evidence |
|
| Contrary Evidence |
|
| Rating Justification | HIGH because: capability confirmed, authority established, intent demonstrated in adjacent domains (tariffs, tech bans). Contrary evidence does not negate threat existence—it suggests execution would be considered escalatory. |
3.2 Scenario B: Intelligence Operation
Evidence for HIGH Probability Rating
| Capability | Confirmed - FISA 702 compels provider cooperation |
|---|---|
| Legal Authority | Confirmed - FISA Section 702 renewed 2024; explicit authority for foreign surveillance |
| Intent Evidence |
|
| Contrary Evidence |
|
| Rating Justification | HIGH (likely already occurring) because: documented historical precedent of allied leader surveillance; legal authority explicitly authorises this; gag orders mean targets never informed. Reforms provide procedural safeguards but do not eliminate capability or authority. |
3.3 Scenario C: Geopolitical Crisis (Service Denial)
Evidence for MEDIUM Probability Rating
| Capability | Confirmed - Centralised control planes enable instant termination |
|---|---|
| Legal Authority | Confirmed - IEEPA and Executive Order authority |
| Intent Evidence |
|
| Contrary Evidence |
|
| Rating Justification | MEDIUM because: capability and authority confirmed, but intent evidence is extrapolated from adversary treatment (Huawei, Russia, Iran), not allied precedent. This scenario is catastrophic if executed but represents significant escalation from current behaviour. Rated medium due to lack of direct precedent. Economic coercion (tariffs) is documented; cloud weaponisation against allies is not. |
3.4 Scenario D: Data Hostage
Evidence for MEDIUM Probability Rating
| Capability | Confirmed - Provider KMS systems enable key revocation |
|---|---|
| Legal Authority | Uncertain - Would likely require IEEPA emergency declaration |
| Intent Evidence |
|
| Contrary Evidence |
|
| Rating Justification | MEDIUM because: technically trivial to execute; legal pathway exists but untested; no direct precedent but adjacent precedents (asset freezes) exist. Rated medium not low because: (1) capability is undeniable, (2) there are plausible legal mechanisms, (3) impact would be catastrophic if executed. |
4. Quantitative Data Points
4.1 CLOUD Act and Data Requests
Available statistics from transparency reports and government sources:
| Data Point | Value | Source |
|---|---|---|
| UK requests under CLOUD Act agreement (to Oct 2024) | 20,142 total | Lawfare |
| UK CLOUD Act results (H1 2024) | 368 arrests, 3.5 tons drugs seized, £5M recovered | Lawfare |
| Microsoft enterprise requests (H2 2024) | 173 total; 38% resulted in disclosure | Microsoft |
| Microsoft content disclosure rate (2024) | 4.94% of all law enforcement requests | Microsoft |
| AWS foreign government data disclosures to US (to June 2025) | Zero reported | AWS |
| Apple US authority device requests (H1 2024) | 12,043 requests; 85% compliance | Apple |
| CLOUD Act executive agreements | 2 (UK and Australia) | CRS |
Note on Data Limitations: Transparency reports have significant gaps. National Security Letters are reported in ranges (e.g., "0-249") due to legal restrictions. FISA orders are similarly obscured. The statistics above represent the visible portion of government data access; actual access is likely higher.
4.2 Market Concentration
| Provider | Q3 2025 Share | Source |
|---|---|---|
| AWS | 29% | Canalys/Synergy Research |
| Microsoft Azure | 20% | Canalys/Synergy Research |
| Google Cloud | 13% | Canalys/Synergy Research |
| Oracle Cloud | 2% | Canalys/Synergy Research |
| Combined US providers | ~67% | Calculated |
5. Limitations, Caveats, and Counter-Arguments
5.1 Acknowledged Limitations
- Advocacy position: This assessment is prepared to support a specific policy position (sovereign cloud investment). While evidence is cited accurately, the selection and framing inherently supports the conclusion.
- Intent projection: Intent is the least certain element. Projecting from tariff behaviour to cloud service termination involves inference.
- No direct precedent: The most severe scenarios (C, D) have no direct historical precedent against allied democracies.
- Classified information: We cannot assess classified threat intelligence. NCSC, DGSI, CSIS, ASD-ACSC may have additional information that changes the assessment.
- Provider pushback: Major providers have legal teams and have resisted some government demands. This resistance capability is not fully modelled.
5.2 Counter-Arguments and Responses
Counter-Argument 1: "AWS says they've never disclosed foreign government data"
Counter-Argument: AWS's transparency report states zero foreign government content disclosures to US government.
Response: This is accurate but incomplete. (1) NSL gag orders may prevent reporting; (2) FISA orders are reported in ranges; (3) past non-disclosure does not guarantee future protection; (4) the legal authority exists regardless of whether it has been exercised.
Counter-Argument 2: "CLOUD Act has judicial oversight"
Counter-Argument: CLOUD Act requires warrants approved by US courts.
Response: US courts apply US law. A foreign government has no standing to challenge a US warrant. Judicial oversight protects US constitutional rights, not foreign sovereignty.
Counter-Argument 3: "Providers would refuse unlawful orders"
Counter-Argument: Major tech companies have legal teams and have challenged government overreach.
Response: Providers can only challenge orders they believe exceed legal authority. IEEPA and FISA provide broad, judicially-validated authority. An order under valid legal authority must be obeyed regardless of corporate preference.
Counter-Argument 4: "This has never happened to an ally"
Counter-Argument: The US has never terminated cloud services to a democratic ally.
Response: Correct. This is why service denial scenarios are rated MEDIUM not HIGH. However: (1) absence of past occurrence does not mean future impossibility; (2) the capability and legal authority are confirmed; (3) the current political environment shows increased willingness to coerce allies.
6. Recommendations for Independent Verification
Readers seeking to verify or challenge this assessment should:
- Request classified briefing: Ask NCSC (UK), ENISA (EU), CCCS (Canada), or ASD-ACSC (Australia) for classified threat assessments on cloud sovereignty.
- Commission independent legal analysis: Instruct counsel specialising in international law and US national security law to assess CLOUD Act, FISA 702, and IEEPA authorities.
- Review provider contracts: Examine termination clauses in existing government cloud contracts with US providers.
- Consult transparency reports directly: Review Microsoft, Google, AWS transparency reports.
- Review Schrems II ruling: The CJEU judgment provides judicial assessment of US surveillance risk.
Methodology Summary
Key transparency: The most severe scenarios lack direct precedent. Ratings reflect capability and authority even where intent evidence is extrapolated. Independent verification is encouraged.