{
  "id": 11219656,
  "title": "Texas HB 149 vs the EU AI Act: What Engineers Must Build",
  "url": "https://urgent.news/2026/10/01/texas-hb-149-vs-the-eu-ai-act-what-engineers-must-build",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-01T15:18:33.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/subodhkc/texas-hb-149-vs-the-eu-ai-act-what-engineers-must-build-26pb"
  },
  "original_language": "en",
  "account": "AI systems do not remain static after compliance measures are implemented. Models change, data sources update, prompts evolve, and autonomy settings adjust. An engineer might introduce new tools, expand the system's applications beyond the initial review scope, or face unforeseen changes that render the previously approved application no longer valid. Consequently, engineering leaders must treat regulations like Texas HB 149 and the EU AI Act as ongoing specifications to ensure the running system remains compliant, rather than static collections of forms to be checked off. Both laws, despite their differences, ultimately demand similar engineering capabilities: maintaining awareness of the deployed AI, understanding the rationale for its operation, tracking changes that occur, accumulating relevant evidence, and ensuring the system remains within its approved boundaries.\n\nTexas HB 149, enacted in January 2026, applies broadly to any business operating in Texas, selling products or services to Texas residents, or developing/deploying AI systems within the state. The EU AI Act, on the other hand, operates in phases, with prohibited practices and AI-literacy provisions taking effect in February 2025, governance rules for general-purpose AI models starting in August 2025, and Article 50 transparency requirements becoming applicable on August 2, 2026.\n\nWhile both regulations aim to ensure AI is used responsibly, they do so in distinct ways. TRAIGA, Texas' AI law, does not follow a risk-tiered approach like the EU AI Act, which categorizes AI systems into prohibited uses, high-risk systems, systems requiring transparency, and those with minimal/no-risk. Instead, TRAIGA merges broad applicability with specific prohibited uses, disclosure requirements, biometric rules, anti-discrimination provisions, government restrictions, an Attorney General enforcement mechanism, a cure process, and liability protections tied to testing and risk-management frameworks.\n\nOne key difference is the lack of an EU-style high-risk or low-risk classification system in Texas. This means that engineering teams should not attempt to force their systems into such classifications, as this does not meet the requirements of Texas law. Another critical distinction is the absence of a universal 60-day cure right in Texas, unlike the EU AI Act, which mandates a 60-day opportunity for rectification after being notified of a violation.\n\nMoreover, TRAIGA's disclosure obligation is narrower than a universal commercial chatbot disclosure requirement. The law specifically mandates disclosure when a governmental agency makes an AI system available to consumers and when AI is used in connection with healthcare services or treatment. In contrast, the EU AI Act's Article 50 includes transparency obligations for certain direct AI interactions and AI-generated or manipulated content, extending beyond the scope of TRAIGA.\n\nFinally, Texas HB 149 does not proactively address AI bias in the abstract. The statute only prohibits the intentional development or deployment of AI systems designed to unlawfully discriminate against a protected class. Disparate impact alone is insufficient to demonstrate intent. An engineering dashboard that claims a fairness metric falls outside a threshold equals a TRAIGA violation would be inventing a legal standard that does not exist in the statute.",
  "summary": "This article is an engineering and governance analysis, not legal advice. Legal applicability and interpretation should be reviewed with qualified counsel. The Engineering Problem Underneath Two Different Laws Production AI does not stand still after a compliance checklist is completed. The model changes. A vendor replaces an underlying model. Retrieval data changes. A prompt changes. An engineer…",
  "key_points": [
    "Texas HB 149 applies to businesses in Texas, regardless of size.",
    "EU AI Act has phased implementation, starting in 2025.",
    "TRAIGA lacks EU-style risk classification system."
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}