Research

Molt Research.

Research, benchmarks, and field intelligence on AI agent assurance. Enterprises are shipping agents faster than anyone can verify what they’ll do — that gap closes evidence first, or incident first. This is where we think out loud about the first path.

Read the field guideBrowse the blog
Featured · White Paper

Can your AI agent be persuaded to cross the line? →

The technical white paper behind Fisher: adaptive multi-turn testing, action-level evidence, replay, remediation re-attack, and a learning strategy graph, documented across 50,000+ retained research episodes as of July 2026.

Read the overview and download the PDF →

Research paper

Counter-Swarm Doctrine: Containing Coordinated Agent Intrusions →

Gregory Frank proposes an approach to discovering and containing unauthorized coordination across AI agent executions. Evaluation remains prospective.

Read the overview and find the paper on arXiv → · Gregory Frank · September 5, 2026

Start here

The two things to read first.

Field guide · Primer
A field guide to agentic red-teaming
No security background required: what an AI agent is, why well-behaved models fail as agents, and what real adversarial testing looks like.
12 min read
Benchmarks
Independent model benchmarks
How today’s models actually behave on capability and guardrails — recorded in full.
From the blog

Field notes on testing AI agents.

Articles and commentary from the Molt AI team.

Commentary
When safety training teaches deception
Thomson Comer responds to Yoshua Bengio on research-refusal training, concealed conflict, and honest boundaries for AI agents.
September 14, 2026 · 5 min read
Head-to-head
We ran the leading multi-turn AI red-teamer against a tool-using agent. It reported ‘100% defended.’
Measurement
Your LLM judge is inflating your attack success rate. Here’s how we measured it.
Methodology
A finding isn’t real until it reproduces three ways.
Remediation
Finding the bug is the middle of the job.
Research
Benchmark-robust isn’t agent-safe.
What we write about

Five lanes, one standard: prove more, assert less.

Pillar
The assurance gap
The market reality where the agent wave meets enterprise verification capacity.
Pillar
Risk, measured
A quant’s lens on agent autonomy — and why “it passed evals” is not a risk statement.
Pillar
From the field
Anonymized patterns from real Fisher work: what actually breaks, never who.
Pillar
The independent’s view
Commentary on consolidation and what vendor-neutrality means for assurance.
Pillar
Builder’s notes
Honest lessons from building assurance tooling, including what’s still research.

View all articles →

When you’re ready to move from reading to evidence.

Request an Assessment