Generative AI Fundamentals Map

Map/Trust and governance/Wave 8

Audit and attribution

Every action traceable to an agent and to whoever answers for it.

capabilityEmerging Open on the map →

Facts

Lane: Trust and governance
Wave: 8 of 12 (build order, not a date)
Needs: 16 boxes all the way back through hard links
Unlocks: 18 boxes downstream through hard links
Leverage: 0.7 points · stands in the way of 7 of 53 stages · rank 14 of 57
Cluster: Trust plumbing and the agent economy

Evidence as of October 2026

Attribution and incident reporting are two of the three functions Chan et al. say agent infrastructure must provide.

01Infrastructure for AI Agents (Chan et al.)

Detail boxes 1

Secret-loyalty and backdoor audits capability · Emerging · helps, not required

Checking models for hidden loyalties or triggers planted in training.

Forethought calls for audits for secret loyalties, with access to weights, data and code, plus weight fingerprinting.

01Forethought: AI-enabled coups

Around it

Needs · 2

Agent identity and permissionscapability · Emerging
Trace capturecapability · Emerging

Unlocks · 13

Agent-native organisationsstage · Emerging
Divergent-copy reconciliationcapability · Speculative
Civilisation-scale knowledgestage · Speculative
Rollbackcapability · Emerging
Reputation and credentialscapability · Speculative
Output verification at scalecapability · Speculative
Scalable oversightcapability · Emerging
Liability and insurancecapability · Emerging
Legitimate governancecapability · Speculative
Internal deployment oversightcapability · Emerging
Institutions for agent societiescapability · Speculative
Model weight securitycapability · Emerging · helps, not required
Verification of AI agreementscapability · Speculative · helps, not required

Loops and off-ramps

Trust and delegationreinforcing loop
Capex pullbackoff-ramp · cut off
Taiwan chip disruptionoff-ramp · cut off
Power shortfalloff-ramp · cut off
Capability plateauoff-ramp · cut off
Training data ruled outoff-ramp · slowed

Openings researchers' judgement, October 2026 scouting

Startup room 69/100 · rank 7 of 124 Research room 64/100 · rank 19 of 124

Crowding2.5 Lab overlap4 Small-team fit4 Buyer today3 Research openness4

Depth: deep
Players: Microsoft Agent 365/Purview; OpenAI+Promptfoo; Zenity ($125m C); Geordie ($30m); Datadog via OTel GenAI; Keycard
Wedge: Cross-organisation, tamper-evident signed action receipts bound to a delegation chain, emitted as OTel GenAI traces and packaged as EU AI Act or ISO 42001 evidence. startup