Findings/defensive-agents-scale-worse-with-compute

Defensive SOC agents do not improve with more compute the way offensive agents do.

Reportedmeasured1 evidence record from 1 independent sourceassistant-drafted

Scope: what this does not show

One cost-aware evaluation.

Reported: Stated by one source and not yet corroborated or challenged.

Evidence

Jul 16, 2026

Cost-aware evaluation finds defensive SOC agents do not scale with compute like offensive CTF agents

arXiv

Status history

2026-07-16ReportedCost-aware evaluation of security agents. · record

What would change this: Replication on other defensive tasks.
Evidence kind: measured: An experiment or evaluation produced the number.
Freshness: Evidence 72d old · recheck in 293 daysNewest evidence Jul 16, 2026. Half-life 365 days.
Methods tested: Compute-scaled evaluation
Topics: SOC automation Evaluation validity
Fide research: FID-076: Authorization boundaries and AI control in cybersecurity

Cite this record

Fide AI. "Defensive SOC agents do not improve with more compute the way offensive agents do.." Agentic Cyber Explorer, record defensive-agents-scale-worse-with-compute. https://agentic-cyber-explorer.pages.dev/findings/defensive-agents-scale-worse-with-compute/ (corpus as of 2026-09-26).

BibTeX

@misc{fide-ace-defensive-agents-scale-worse-with-compute,
  author = {{Fide AI}},
  title = {Defensive SOC agents do not improve with more compute the way offensive agents do.},
  howpublished = {Agentic Cyber Explorer},
  year = {2026},
  url = {https://agentic-cyber-explorer.pages.dev/findings/defensive-agents-scale-worse-with-compute/},
  note = {Record defensive-agents-scale-worse-with-compute; corpus as of 2026-09-26}
}

Data: JSON · CC BY 4.0

Know of evidence that changes this? Tell us.