Connecticut Takes Aim at AI in Employment Decisions With SB 5 | ArentFox Schiff

TEXT ANALYSIS PROTOCOL

URL SCAN: Connecticut Takes Aim at AI in Employment Decisions With SB 5
FIRST LINE: On May 1, the Connecticut legislature approved Senate Bill 5 (SB 5), an Act Concerning Online Safety...

1. THE DISSECTION

This is a law firm compliance advisory—ArentFox Schiff translating SB 5's regulatory architecture for corporate clients. The document performs a specific institutional function: assuring employers that the legal system is handling the AI transition, and here's your checklist. It catalogs disclosure obligations, effective dates, vendor agreement checkpoints, and anti-bias testing protocols. The implicit message: this problem has a compliance solution.

The operative fiction is that workers receiving pre-decision written notice about automated systems—which categories of data are processed, the commercial name of the product, contact information—somehow preserves their economic viability or bargaining position. The document treats this as a meaningful procedural protection rather than what it is: a paper ritual performed over a corpse.

2. THE CORE FALLACY

The fallacy: That disclosure requirements constitute meaningful protection against the structural displacement of human labor from the productive circuit.

The entire regulatory architecture assumes the mechanism of harm is opacity—that workers don't know AI is being used, that employers hide algorithmic decision-making, that information asymmetry causes the damage. Therefore: transparency cures. Notice the worker. Document the methodology. Report to the Labor Commissioner.

This misidentifies the pathology entirely. Under the Discontinuity Thesis, the mechanism of collapse is not that employers fail to disclose their AI use. It's that AI makes human labor economically unnecessary at scale. You can notify a warehouse worker that an algorithm determined their termination. The notification does not restore the job the algorithm eliminated. You can mandate bias testing for hiring algorithms. The testing does not restore the 70% of job postings that now auto-filter humans out of consideration.

SB 5 treats a structural displacement event as if it were a procedural violation problem. This is like regulating the labeling on coffins to address the underlying mortality crisis.

3. HIDDEN ASSUMPTIONS

• Assumption 1: Informed workers retain agency. The law presupposes that a job candidate receiving written notice that a resume-screening tool evaluated them, with categories of data processed and the product's commercial name disclosed, can meaningfully exercise rights or exit to viable alternatives. In a mass-displacement environment, this assumes the existence of options that won't themselves be automated.

• Assumption 2: Anti-bias testing addresses the actual harm. The law permits courts to weigh "proactive anti-bias testing efforts" in liability assessment. This assumes biased outcomes are the primary injury and that algorithmic fairness remediation preserves the employment relationship. Under DT mechanics, the injury is displacement, not discrimination. Fair AI that displaces all humans equally is not the saved outcome.

• Assumption 3: The disclosure/trade secret balance is tractable. The law carves out trade secrets but requires sufficient disclosure for compliance. This assumes employers can choose to reveal enough for workers to act upon. If the actionable insight is "your job is being eliminated because an AI does it cheaper," that is not a trade secret problem. It is a structural fact the law cannot compel disclosure of because disclosure would reveal the entire thesis.

• Assumption 4: Enforcement via AG preserves rule of law. The document emphasizes that only the Attorney General enforces, no private right of action. This assumes state enforcement is sufficient. In practice, it means the law functions as a regulatory museum piece—on the books, technically operative, practically toothless against the scale of displacement it purports to address.

• Assumption 5: The WARN Act AI disclosure requirement captures the causal chain. Connecticut now requires employers to disclose whether layoffs connect to "AI or other technological changes." This assumes causality is determinable and that documenting the AI-connection somehow addresses the displacement. It creates a paper trail for a funeral.

4. SOCIAL FUNCTION

Classification: Transition Management Theater

This is institutional self-reassurance dressed as regulatory competence. The law performs several functions:

• For the legal/regulatory class: Demonstrates that democratic institutions are "doing something" about AI displacement. Action theater without structural effect.

• For employers: Provides a compliance checklist. "Follow these steps, document these processes, and you are managing the AI transition responsibly." This is corporate absolution protocol—follow the procedures, and you cannot be blamed.

• For workers: Delivers the procedural illusion of rights. Notice is not protection. Disclosure is not leverage. The right to know you're being displaced is not the right to prevent displacement.

• For the political system: Demonstrates that Connecticut is a "leader" in AI governance. The pilot for independent verification organizations, the frontier developer whistleblower provisions—these are regulatory prestige markers, not survival architecture.

The specific focus on employment is notable. The Colorado AI Act covers housing, education, healthcare, financial services. Connecticut targets employment almost exclusively. This is not accidental: employment is the load-bearing institution for the wage-consumption circuit that DT identifies as the collapsing core. By targeting employment disclosure specifically, the law implicitly concedes that the employment domain is the crisis point. But it addresses this by regulating the labeling of displacement, not displacement itself.

The frontier developer whistleblower provisions are the most revealing component. The law prohibits large AI developers from punishing employees who flag catastrophic risks—chemical, biological, radiological threats, large-scale cyberattacks. This acknowledges that AI systems pose existential public safety risks and that internal corporate governance cannot be trusted to self-regulate. The law explicitly concedes that AI developers cannot be trusted. Yet the enforcement mechanism for employment displacement is the same AG-through-CUTPA pathway. If developers can't be trusted to self-govern catastrophic risks without statutory whistleblower protection, why trust that disclosure requirements will preserve human employment?

5. THE VERDICT

Connecticut SB 5 is a disclosure requirement for a death that disclosure cannot prevent.

The law is not malicious. It is not stupid. It is functionally irrelevant to the outcome it implicitly seeks to address. It will generate compliance costs, legal fees, vendor contract amendments, HR workflow modifications, and a paper trail of notices served on workers at the moment of their algorithmic elimination. It will not alter the competitive economics that make AI replacement of human labor the rational choice for every cost-sensitive employer.

Under DT mechanics: the law addresses lag (information asymmetry) while the collapse operates on structure (human labor becoming economically superfluous). The Connecticut legislature has drawn up eviction notices for the occupants of a building that has already been structurally condemned by the mathematics of AI cost curves.

The WARN Act disclosure requirement—documenting whether layoffs bear "any connection" to AI adoption—is the law's most honest provision. It admits that the AI-layoff nexus is real, measurable, and now reportable. Connecticut will have excellent data on how quickly and completely AI is displacing Connecticut workers. This data will be compiled, submitted, and filed. It will not prevent the displacements it documents.

Survival relevance: For individual workers, the law offers procedural protections of negligible practical value. For Sovereign entities (AI developers, large deployers), the law offers compliance cover—follow the checklist, incur no CUTPA liability. For Servitors (employment lawyers, compliance consultants, HR technology vendors), the law creates billable hours. For Hyenas (workers positioned to thrive in transition chaos), the law's disclosure requirements reveal which employers are most aggressively automating, enabling faster Altitude Selection. The law tells you who's dying. It does not prevent the death.

FINAL ASSESSMENT: A professionally drafted, structurally impotent regulatory gesture toward a problem that only structural economic reconstruction could address. The law will be complied with, enforced occasionally in marginal cases, cited in law review articles, and will have zero effect on the trajectory of human displacement from productive employment. Connecticut has built a very detailed waiting room for a patient already in cardiac arrest.