Free Monthly Digest

Slop Tracker

AI-generated code is ~42% of committed code. It ships 2.7x more vulnerabilities. Insurers are excluding it. Governments are fining it. This newsletter tracks what matters.

42%

of committed code is now AI-generated

Sonar 2026, n=1,100+

2.74x

more vulnerabilities in AI code vs human

Veracode 2025

19%

slower with AI tools (devs think 20% faster)

METR RCT, arxiv 2507.09089

21%

of enterprises have mature AI governance

Deloitte 2026

What you get each month

Incident card — one named production incident, dissected: what happened, who caught it, what governance primitive would have prevented it
By the numbers — 5-7 hard stats with primary-source citations you can drop into a board deck
Regulatory watch — EU AI Act enforcement, US state patchwork, insurance exclusions, case law (Moffatt, Mobley, Berkley)
The wedge — what the governance vendor landscape is missing and where the market is heading
Quote of the month — one positioning line backed by evidence

Issue #1 Preview

HUMMBL Slop Tracker — Issue #1 April 2026

5 things you should know:

AI code is 42% of commits and ships 2.74x more vulnerabilities — the security pass rate has been stuck at ~55% for 2 years despite model improvements
SWE-bench is inverted: Claude Opus 4.6 hits 79.3% Verified but produces vulnerable code in 29.2% of samples
Berkley wrote absolute AI exclusions into D&O and E&O policies — no governance = no coverage
Moffatt v. Air Canada (2024) + Mobley v. Workday (cert. July 2025) are settling the liability chain
EU AI Act enforcement goes live August 2, 2026 — Finland has been enforcing since January

Full issue: ~1,500 words, 7 cited stats, 3 court cases, 1 incident card. Evidence corpus on GitHub.

Who reads this

CISOs evaluating AI code risk. CAIOs building governance programs. GCs tracking AI liability exposure. Engineering leaders who need cited evidence, not vibes, for their next board presentation.

Want a governance posture check?

Take the 20-question self-assessment. Get a scored report mapped to NIST AI RMF and ISO 42001.

Start Assessment