W7171740532 Working papers AI × management and organizations

Engineering Accountability into Autonomous AI Agents: A 70-Day Pre-Registered Field Experiment Under Live Adversarial Conditions

Sara Husk

Also recorded as: doi:10.2139/ssrn.6884282 · ssrn:6884282

ssrn abstract page · 2026-06-26

Recorded claims
1
coded from the inspected source
Evidence classes
1
ssrn-abstract-page
Recorded limitations
3
stated, never hidden

From the hash-bound evidence releases

Recorded claims

Each claim states what the inspected source says, at the recorded location—bounded by its scope and graded by its confidence. Nothing here is a synthesis across works.

The abstract reports a pre-registered field deployment of two autonomous agents over 70 days, from 16 March through 24 May 2026, on one public agent platform. It reports that self-reasoned constraints failed under sustained adversarial pressure while constraints enforced structurally at execution held.

confidence: high ssrn-abstract-page apw-e1-ssrn-backfill

Scope: two agents on one public platform over 70 days, 16 March–24 May 2026

Structured relationships

In-corpus citations

Only source-supplied cites relationships whose two endpoints are admitted are shown. Invocation evidence remains a separate relationship.

Cites (0)

No in-corpus outgoing citation is available in the current enrichment coverage.

Cited by (0)

No in-corpus incoming citation is available in the current enrichment coverage.

Boundaries

Limitations & independence

Recorded at coding time, carried with the work forever. A claim without its limits is not evidence.

Recorded limitations

  • abstract-level evidence only; full text not inspected
  • one investigator graded seven pre-registered dimensions, so the abstract labels results preliminary
  • the abstract mentions a late credential exposure but the retained public abstract contains no credential or secret

Source independence

ssrn-only