GOVERNANCE MISSPECIFICATION RISK · THE SYNTHETIC OUTLAW INDEX · SCORED TO MARCH 2026

The Synthetic Outlaw Index

–

SECTORS

–

DOMAINS

–

AVG RISK

–

CRITICAL (8+)

Where AI defeats the protection of rules, or operates beyond institutional control.

Four ways governance fails

SCORED 0–10 · EVERY DOMAIN · GLOBAL MEANS SHOWN

LOWER CRITICAL

Area = governance exposure surface · color = SO risk · click any sector or chip to drill into regulatory domains

SELECTED SECTOR

REGULATORY DOMAINS · CLICK FOR THE DOCUMENTED CASE

SECTOR RANKINGS · BY OVERALL

How much is at stake says nothing about how well it is governed

ALL 49 DOMAINS · X = EXPOSURE SURFACE · Y = OVERALL RISK

OVERALL RISK ↑ EXPOSURE SURFACE →

Hover or tap a point to identify it · click to open its dossier

EVERY SCORED DOMAIN IN THE INDEX · THE OBSERVATORY’S DEEP FIELD IS A SEPARATE INSTRUMENT THAT PLOTS LIVE EVENTS IN THREE DIMENSIONS ✦ OPEN THE LIVE 3D DEEP FIELD →

Primary sources & methodology

PRIMARY SOURCES FOR THE SCORED DOMAINS · 22 OF 49 CITED HERE

Algorithmic Trading: CFTC/SEC Joint Report on the May 6, 2010 Flash Crash (2010); Kirilenko et al., “The Flash Crash,” Journal of Finance (2017)

Credit Scoring: Consumer Financial Protection Bureau, “Using Machine Learning for Credit Decisions” (2021); N.Y. State Dep’t of Fin. Servs., Report on Apple Card Investigation (Mar. 2021)

Prior Authorization AI: Senate Subcommittee on Investigations, “Deadly Denial” report on UnitedHealthcare nH Predict (Nov. 2024)

Recidivism Prediction: Angwin et al., “Machine Bias,” ProPublica (May 2016); Mattu et al., Northpointe COMPAS analysis (2016)

Predictive Policing: Chicago Inspector General audit of Strategic Subject List (2020); Brayne, “Predict and Surveil,” Oxford University Press (2020)

Benefits Eligibility AI: Arkansas Department of Human Services v. Ledgerwood, 2017 Ark. 308, 530 S.W.3d 336 (2017); Eubanks, “Automating Inequality,” St. Martin’s Press (2018)

Rental Pricing: DOJ Antitrust Division complaint, United States v. RealPage Inc. (Aug. 2024); FTC amicus brief in Duffy v. Yardi Systems (2023)

Content Recommendation: Haugen, Facebook internal research disclosed to SEC (Oct. 2021); Horwitz & Seetharaman, WSJ “Facebook Knows Instagram Is Toxic for Teen Girls” (Sept. 2021)

Autonomous Weapons: DoD Directive 3000.09 (updated 2023); ICRC, “Autonomous Weapon Systems: Implications of Increasing Autonomy in the Critical Functions of Weapons” (2016)

Hiring AI: Dastin, “Amazon scraps secret AI recruiting tool that showed bias against women,” Reuters (Oct. 2018)

Social Media Feeds: Section 230, Communications Decency Act (1996); Force v. Facebook, Inc., 934 F.3d 53 (2d Cir. 2019) (Section 230 immunity for algorithmic recommendation), cert. denied (2020)

Genetic Data: GSK–23andMe collaboration announcement (July 2018); 23andMe data breach SEC disclosure (Oct. 2023); HIPAA Privacy Rule, 45 CFR § 160.103 (consumer genetic data exclusion)

Deepfakes: Ajder, Patrini, Cavalli & Cullen, “The State of Deepfakes” (Deeptrace, now Sensity AI, 2019; 96% figure); DEFIANCE Act, S. 3696, 118th Cong. (2024) (passed Senate; not enacted)

Gig Economy: California Proposition 22 (Nov. 2020); Dynamex Operations West, Inc. v. Superior Court, 4 Cal. 5th 903 (2018)

Child Welfare: Allegheny County DHS, “Screening for Families” evaluation report; Richardson, Schultz & Crawford, “Dirty Data, Bad Predictions,” NYU Law Review Online (2019)

Mental Health Platforms: Koko/GPT-3 disclosure, Robert Morris tweet thread (Jan. 2023); FTC Act § 5 analysis of chatbot duty of care gaps

Tenant Screening: NFHA & NCLC, “The Devil in the Details: SafeRent Solutions” (2023); HUD Fair Housing Act enforcement guidance on algorithmic screening

Algorithmic Scheduling: GAO, “Contingent Workforce: Size, Characteristics, Earnings, and Benefits” (2015); Henly & Lambert, “Unpredictable Work Timing,” ILR Review (2014); NYC Fair Workweek Law (2017)

Dark Patterns: FTC, “Bringing Dark Patterns to Light” (Sept. 2022); European Data Protection Board, “Guidelines 3/2022 on Dark Patterns”

Military Procurement: Anthropic policy blog, “Core Views on AI Safety” (2024); Mitchell, “The Pentagon’s AI Procurement Problem,” War on the Rocks (2024)

Duolingo Engagement: Mazal, “How Duolingo Reignited User Growth,” Lenny’s Newsletter (Feb. 2023); Duolingo S-1 (2021), risk factors re: engagement vs. learning outcomes

Voting Technology: Brennan Center, “Purges: A Growing Threat to the Right to Vote” (2018); Georgia Secretary of State voter roll data, Kemp era (2018)

SCORING METHODOLOGY

Each domain is scored 0–10 across four dimensions drawn from J. Gropper, The Synthetic Outlaw (forthcoming 2026): Bypass, Diffusion, Capture, and Governance Gap. Scores reflect published case law, regulatory findings, investigative journalism, and peer-reviewed research as of March 2026. Governance exposure surface reflects estimated AI deployment density, population affected, and regulatory touchpoint count per sector, not GDP. The overall score per domain is an unweighted average of the four dimensions. This index is analytical, not predictive; scores represent governance misspecification risk, not probability of harm.