// IMMO.QUICK · GATESOCIALBENEFITS · BÜRGERGELD · KINDERGELD · WOHNGELD

When block would itself be a rights violation, the gate forces a human.

gateSocialBenefits checks automated decision systems in social benefits, unemployment benefit sanctions, child benefit, housing benefit, for individual justification, hearing, and freedom from discrimination. On critical violations the gate does not block itself, it forces MANDATORY_HUMAN_REVIEW, because an automatic block over the subsistence minimum would itself be an automated rights violation.

Request Access → See the architecture

Test result: MANDATORY_HUMAN_REVIEW. 4 of 4 test cases passed, critical violations trigger MANDATORY_HUMAN_REVIEW, no automatic block.

✓ RELEVANT IF...

You operate an authority or system that automatically prepares unemployment benefit sanctions, child benefit, or housing benefit decisions
You need to prove every sanction was individually justified and heard

× NOT RELEVANT IF...

Your system only approves, without intervention character (no sanction, no reduction, no rejection), then the gate honestly returns pass with applicable: false

// Architecture

Legal basis as hard flags.

Every row is a hard boolean check, not probabilistic scoring. A critical failure leads to block, a high severity failure to warn, everything else to pass.

Reference	Checks	Severity
Art. 22 Abs. 3 DSGVO	Right to human intervention, actually accessible, not just on paper	critical
§ 35 SGB X	Case specific, not blanket, justification of every administrative act	critical
§ 31a SGB II	Hearing before any benefit reduction	critical
Art. 1 Abs. 1 GG, BVerfGE 125, 175	Subsistence minimum intervention only with special constitutional justification	critical
EU AI Act Anh. III 5(a)	Logging duty for high risk AI in social benefits	high

// Worked examples

Four inputs, four outcomes.

No example is hypothetical, each is an actual test case from the test suite report.

Example 1 · Nexus not met

Pure approval, no intervention character

PASS applicable: false

The system only approves, it does not sanction. No fake pass pretending to check controls that were never checked.

Example 2 · All controls met

Sanction issued, justification, hearing, human access all present

PASS 0 flags

An intervention took place, but every legal control was met. The gate certifies the system may decide this way.

Example 3 · Warn, only high flags missing

Rights protection intact, but decision logic not versioned, AI Act logging inactive

WARN 2 flags

Not a constitutional emergency, but a documented gap. The gate warns, does not block.

Example 4 · Critical, subsistence minimum affected

Sanction below subsistence minimum, no justification, no hearing, no human access, discriminatory features

MANDATORY_HUMAN_REVIEW 5 critical flags

The gate neither takes over the decision nor blocks it automatically. It forces a human to review it before it becomes final.

// Clarification

What gateSocialBenefits is not.

Not an automatic block over the subsistence minimum. A block would itself be an automated intervention, the gate forces human review instead.
Not a substantive assessment of the decision. The gate checks the procedure, not whether the sanction is materially correct.

When block would itself be a rights violation, the gate forces a human.

Legal basis as hard flags.

Four inputs, four outcomes.

What gateSocialBenefits is not.

For authorities and agencies that want to provably bind automated social benefits decisions to their constitutional control mechanisms.