Bangladesh · US — an assembled comparison from each country's own governance data.
Bangladesh and US illustrate different points on the AI governance spectrum. Bangladesh's approach — Draft risk-based national AI policy atop a newly enacted data-protection law, carried across a 2026 change of government; US's approach — Sectoral and executive-led, with significant state-level activity.
On trajectory: Bangladesh is tightening its regulatory grip; US is loosening its regulatory grip. Most recent tracked marker for each: Bangladesh — "The Personal Data Protection Act, 2026 comes into force (reported around 15 April 2026), replacing the 2025 Ordinance to become…" (2026-04-15); US — "Newsom signed a package of child-online-safety bills, several aimed directly at AI." (2026-09-10).
A risk-based National AI Policy 2026-2030 still in draft, sitting atop a newly enacted Personal Data Protection Act 2026 and a 2025 Cyber Security Ordinance
Sectoral, executive-led — no binding federal frontier-model law
Bangladesh (tightening): Bangladesh appears to be moving from having essentially no dedicated AI or data-governance framework toward one: a first comprehensive data-protection law is now in force, a reformed cyber ordinance has replaced the 2023 Act, and a risk-based national AI policy sits in advanced draft. The direction is therefore toward more formal regulation, but implementation is genuinely uncertain — the coordinating AI agency the policy envisages does not yet exist, the AI policy is unadopted, and the whole agenda has just passed from an interim government to a newly elected one whose priorities are still being set out. Its clearest legislative marker on file is "Personal Data Protection Act, 2026" (in force). US (loosening): The federal posture remains deregulatory on consumer-facing and precautionary rules, but the direction is increasingly asymmetric: on national-security grounds, EO 14409 ('Gold Eagle') and the July agentic-hacking disclosures have pushed the government to assert markedly more control over frontier-model access and cyber-capability testing. With the AI & Crypto Czar role vacant since March, CAISI's standards and testing work remains the most consistent thread of federal activity. Its clearest legislative marker on file is "California SB 53 — Transparency in Frontier Artificial Intelligence Act (TFAIA)" (in force).
Leans on a very large IT-freelancer/BPO workforce and 'Digital/Smart Bangladesh' branding, with nascent Bengali-language-model and data-centre ambitions
Infrastructure- and export-led; the 'Stargate' initiative and America's AI Action Plan prioritise compute build-out
Bangladesh: Moderate — the ICT Division and a2i drive digital-government and AI policy, but a dedicated AI regulator (the proposed NDGIA) does not yet exist and the national AI policy remains a draft (notable named developers: Hishab Technologies, Intelsense AI). US: Low-to-moderate — light federal touch, no binding frontier-model law (notable named developers: OpenAI, Anthropic; AWS leads domestic cloud share at ~28%).
UNESCO AI-readiness assessment and long-running UNDP support for a2i; the draft policy signals an intention to accede to the Council of Europe Framework Convention on AI
Chip export-control coordination with allies; no formal multilateral AI safety treaty
Both have live "security" debates on file (Bangladesh: "Data-protection and cyber laws that critics argue protect the state more than citizens"; US: "Agentic AI, cyber-offence and developer accountability"), suggesting overlapping strategic anxieties rather than fully independent postures.
Finalising and implementing the National AI Policy 2026-2030 while operationalising the Personal Data Protection Act 2026
Frontier-model competitiveness and compute infrastructure appear to be higher near-term federal priorities than precautionary regulation, though states continue to pursue their own rules
Where each government is actually spending its attention right now — not a statement of law, but of priority.
4 tracked measures (2 proposed, 2 in force)
7 tracked measures (4 in force, 2 passed, 1 proposed)
Bangladesh tracks 6 government AI institutions; US tracks 3 government AI institutions. Raw counts aren't a maturity score by themselves — a single binding statute can matter more than five proposals — but they mark how much formal apparatus each country has stood up so far.
Frontier-model developers, Dedicated AI regulator/institute, Adopted AI-specific law, Compute and hardware
Federal AI safety statute
The gaps recorded for each country don't overlap — each is missing something different, which is a weaker signal than a shared gap but still worth reading as "not present as of the most recent research," not "definitively absent forever."
Shared
Bangladesh
US
Bangladesh and US each describe their own state's role differently: Bangladesh — "Moderate — the ICT Division and a2i drive digital-government and AI policy, but a dedicated AI…"; US — "Low-to-moderate — light federal touch, no binding frontier-model law". Layered onto trajectory (Bangladesh tightening, US loosening), the open question this comparison actually poses is how much of each country's AI trajectory is being steered by the state versus left to market and research actors — and, for whichever side is tightening, whether that steering is increasing or easing right now.