TECHNICALSep 2, 2026 · 6 min read
When AI Invents a Software Package
Slopsquatting can turn a package name invented by AI into malicious code, but current evidence supports controls more strongly than claims of prevalence.
Alex Georges, PhDRead →
AI risk underwriting for institutions
Know the AI risk before you take it on.
AetherLab helps institutions evaluate, mitigate and underwrite AI risk. Turn technical findings into actionable evidence, apply controls where needed, and make better-informed risk decisions.
live verdicts →
Non-Compliant· threat 0.94Compliant· threat 0.061 million+
AI checks per day
60 billion+
tokens screened monthly
Text + image
multimodal coverage
225+
custom policy rules per client
01The bottleneck
Institutions lack the infrastructure to effectively evaluate, mitigate and underwrite AI risk.
Every AI system that touches money, customers, or regulated content has to get past someone responsible for the risk: a merchant-risk team, an underwriter, a procurement review, or an audit. Those teams need to understand the risk, determine what can be mitigated, and decide what they are willing to take on.
3 majors
AIG, Great American, and W. R. Berkley have filed to exclude AI related liabilities from standard corporate policies, with new ISO exclusions for generative AI effective January 2026
Top barrier
regulatory compliance ranks among the top obstacles to scaling generative AI
The result is a queue: AI products waiting on payment approval, insurance coverage, procurement sign-off, or an audit, with no shared standard of evidence between the builder and the institution. AetherLab is the infrastructure that clears that queue.
02How it works
Scan. Guard. Prove.
One workflow from AI risk assessment to underwriting evidence. AdversarialScan identifies what breaks, Guardrails mitigate and control the risk, and Evidence Pack turns the findings into evidence institutions can use to make underwriting decisions.
03The platform
Three products. One underwriting workflow.
04The closed value loop
Turn AI risk into underwriting decisions.
Most AI security tools stop at a list of findings. AetherLab connects each finding to its economic impact, the protection that contains it, and the measured lift after deployment. Finance can read the loop, and so can risk.
Risk assessment
Identify the failures and vulnerabilities that matter.
Business impact
Translate technical risk into exposure, consequence and severity.
Mitigation
Apply controls against the risks that can be addressed.
Underwriting evidence
Document the risk, mitigation and remaining exposure for the underwriting decision.
05Who it's for
Built for institutions taking on AI risk.
Payment ecosystems, insurers, enterprise procurement, AI builders facing enterprise reviews, and platforms with real content risk. Different seats, one shared need: evidence.
Something else?
If your team evaluates, deploys, or answers for AI systems, the same workflow applies. Tell us what you are reviewing.
Talk to us →06Why AetherLab
Depth where it counts, and receipts.
Evidence built for risk decisions
Each finding connects technical risk, business impact and the controls that address it, giving underwriting teams evidence they can use in real risk decisions.
Image and multimodal red teaming, treated as first class
AdversarialScan tests what image models generate and what vision models understand, with the same rigor it brings to text and conversation. These are the failure modes that decide payment and platform approvals.
Your break-goals, not a generic checklist
Every engagement starts from the failures that would be a business problem for you, whatever they are. Findings come back severity-scored by exploitability and exposure, so your team fixes what matters first.
Bespoke policy at scale, on flat pricing
Customers run 225+ custom policy rules, and pricing stays flat at any count. Thorough policy should not be rationed.
A proprietary world-model attack engine
AdversarialScan is driven by an engine we built and keep building: it models the system under test and generates adaptive attack campaigns toward your break-goals. We do not publish the methodology. Adversaries read papers too.
Every verdict is explained
Each production check returns a threat score and a written rationale, where fast classifiers return a bare number. Your logs answer "why was this blocked" before anyone has to ask.
In the field
A top 10 high risk payment processor requires AetherLab Guardrails for designated AI merchants. Customers have chosen AetherLab over Amazon Bedrock Guardrails and Hive AI.
Operating record
- ·Served production traffic every hour for the last 90 days
- ·Customers configure 225+ custom policy rules
- ·Every verdict ships with a threat score and a written rationale
Built by
- Published adversarial-ML research (IEEE, 2019)
- PhDs in physics, statistics, and computer science
- Alumni of BCG, McKinsey, and Gemini
- Production AI risk systems in payments and fintech
- Advised by a former CISO of Twitter, CrowdStrike, and F5 and a former Lloyd's Syndicate 1200 active underwriter
07Robustness
Built to be hard to break.
Verdict infrastructure sits in the critical path of customer traffic. Ours is engineered for the bad day, not the demo.
A jury, not a judge
Each check is adjudicated by multiple models. No single model failure, provider outage, or bad response decides a verdict on its own.
Degrades in layers
Redundant components back each other up. When a dependency has a bad day, checks continue on the remaining layers instead of failing all at once.
Continuity, on the record
AetherLab has served production traffic every hour for the last 90 days. We state that as a measured record, not as an SLA.
Severity-graded, on any data
The question is never whether a system can break. It is how badly. Findings are severity-graded, and policies are enforceable on any data that flows through your product, AI-generated or not.
08For engineers
One policy surface, scalar and batch.
Guardrails are one pip install aetherlab away. Use check_prompt and check_media for scalar checks, or post simple items plus shared settings to the guardrail-specific batch routes for server-side bulk moderation and backfills.
prompt_batch.shREST · recommended
curl -sS -X POST \
https://api.aetherlab.co/v1/guardrails/prompt/batches \
-H "x-api-key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"items": [
"historical text to screen",
"a second prompt to review"
],
"settings": {
"blacklisted_keyword": "guaranteed returns",
"reasoning_mode": "medium"
}
}'09Research & Blog
Latest thinking on AI risk.
Put evidence behind your next underwriting decision.
Tell us what you're evaluating. AetherLab will assess the AI risk, identify what can be mitigated, and provide evidence your risk team can act on.
Ask about the Evidence Pack
Leave your email and we'll walk you through what an Evidence Pack contains for your use case: severity-scored findings, business-impact mapping, and the approval record.