Mechanism
When AI is in the decision loop, 'who decided' becomes structurally blurry. The lawyer signed, so the lawyer is accountable; the AI suggested the analysis, so the vendor is implicated; the function approved the AI, so the function is accountable; the board approved the programme, so the board is implicated. The blurriness itself, independent of any specific failure, is the risk.
Evidence (what the Evidence Register holds)
One named individual accountable for AI overall, with documented mandate and articulable authority; committee charter; quarterly attestation log; per-Lifecycle-stage accountability assignment.
Mitigation
One named individual accountable for AI overall, with documented mandate and articulable authority; clear delegation chains; documented Lifecycle-stage accountability assignments; per-capability named owner.
Editorial Framing
Accountability dilution is the class that does not require a specific failure to manifest. The blurriness is the risk. The mitigation is one named individual accountable for AI overall, with a documented mandate and articulable authority — and then a clear delegation chain from that individual to each capability's named owner.
Indicative Examples
- Lawyer signed but AI suggested the analysis — who is accountable for the legal advice?
- Function approved the AI but the board approved the programme — who is accountable when something goes wrong?
- Multiple vendors in the AI BoM — who is accountable for a failure that crosses vendor boundaries?