Global Tech Policy Tracker — 2026-04-05

Top Stories

1. EU's AI Act Delays Let High-Risk Systems Evade Oversight

• Region: European Union
• What happened: The EU has delayed key rules in its AI Act, including safeguards for high-risk systems, a ban on AI "nudifier" applications, and sector-specific guidance. Critics argue the delays create a window for high-risk AI deployments to operate without meaningful oversight, undermining the intent of what was billed as the world's most comprehensive AI regulation framework.
• Who's affected: Companies deploying high-risk AI systems across healthcare, employment, and critical infrastructure; civil society groups; and individuals vulnerable to AI nudification tools.
• What's next: The European Parliament has agreed to proposals to simplify AI rules and clarify application dates for high-risk system requirements. Industry observers are watching whether the revised timeline will actually tighten enforcement or further erode protections.

2. Amnesty International Warns EU "Simplification" Laws Will Roll Back Digital Rights

• Region: European Union
• What happened: Amnesty International published a sharp critique of EU proposals to "simplify" technology laws, arguing the changes will roll back fundamental rights protections in order to feed the AI industry's data appetite. The report warns that deregulation framed as administrative streamlining could undermine privacy, expression, and accountability online.
• Who's affected: EU residents, platform users, civil society organizations, and tech companies that have built compliance programs around current rules.
• What's next: The EU's Digital Omnibus package — which includes the simplification proposals — still faces votes from European member state governments and the Parliament. Rights groups are lobbying for amendments to preserve core protections.

1 sources

amnesty.org

amnesty.org

3. China's AI Models Surge to 30% of Global Workloads, Prompting Security Policy Debate

• Region: United States / Global
• What happened: A new analysis published by War on the Rocks reveals that Chinese AI models — including Alibaba's Qwen — have surged from just 1% of global AI workloads in late 2024 to approximately 30% by end of 2025. The analysis argues this rapid spread creates significant national security risks and calls for coordinated policy responses among allied democracies.
• Who's affected: Enterprises and developers globally using Chinese-origin AI models; U.S. and allied government policymakers; national security agencies.
• What's next: The article calls for concrete policy steps to address security risks from Chinese AI adoption, potentially including disclosure requirements, security audits, or restrictions on government procurement.

1 sources

warontherocks.com

warontherocks.com

Regulatory Actions & Enforcement

EU AI Act — European Parliament Approves Simplified Rules and Delayed Timelines The European Parliament's Industry, Research and Energy Committee (ITRE) voted to agree on proposals that simplify AI Act rules and set clearer application dates for high-risk system requirements. The package also includes a ban on AI "nudifier" systems — software that generates non-consensual synthetic nude imagery. However, critics note the delays in enforcement timelines mean high-risk AI systems will continue operating without full oversight obligations for an extended period. The Parliament's agreement follows a Council position adopted in March 2026 that also sought to streamline AI rules.

EU Council Position on AI Streamlining The Council of the EU agreed a position in March 2026 to streamline artificial intelligence rules as part of wider "Digital Omnibus" simplification efforts. While framed as reducing administrative burden, legal observers warn the reforms could weaken accountability mechanisms that the original AI Act was designed to establish. The Council's position now forms the basis of trilogue negotiations with the Parliament.

Industry Response

Baker McKenzie Warns Enterprises on EU Electronic Evidence Rules Law firm Baker McKenzie published guidance warning that EU-wide collection of electronic evidence by enforcement authorities will soon be possible, meaning service providers operating in Europe should begin preparing compliance frameworks now. The firm noted that the expanded enforcement capability — separate from the AI Act — significantly raises the stakes for technology companies operating across EU member states.

Enterprise AI Act Compliance Gap Remains Wide A new readiness report from Vision Compliance (published this week via National Law Review) found that 78% of enterprises are unprepared for EU AI Act obligations. The analysis, based on compliance assessments across eight industries, identified critical gaps in AI system governance structures. The finding underscores an urgent commercial and legal challenge: even as the EU delays some deadlines, companies are far from ready to meet existing ones.

Expert Analysis

• Rights vs. simplification tension deepens in Brussels: Amnesty International's warning that EU simplification laws will "roll back rights" to "feed AI" reflects a growing tension in European tech governance — the same political institutions that championed the AI Act are now under intense industry pressure to reduce compliance costs. Policy researchers should watch whether the Omnibus package weakens enforcement mechanisms or merely streamlines paperwork.

• High-risk AI systems remain in a regulatory gray zone: The combination of delayed AI Act timelines and EU Council streamlining proposals means systems used in hiring, credit scoring, healthcare triage, and law enforcement — all designated "high-risk" — continue operating with fewer guardrails than lawmakers originally intended. TechPolicy.Press analysis notes this creates compounding accountability deficits as real-world harms accumulate faster than enforcement catches up.

• Chinese AI model proliferation demands allied coordination: The War on the Rocks analysis argues that the surge of Chinese AI models to 30% of global workloads is not merely a commercial trend but a national security development requiring coordinated response from democratic allies. The authors suggest disclosure mandates and security review frameworks modeled on existing semiconductor export controls as potential policy tools.

Global Activity Snapshot

What to Watch Next

1. EU Trilogue on AI Act Simplification — Negotiations between the European Parliament and Council will determine how much the Digital Omnibus package weakens or preserves the original AI Act's high-risk safeguards. No fixed date set, but proceedings are actively underway as of early April 2026.

2. EU Nudifier App Ban Implementation — The European Parliament's agreement to ban AI nudification systems still needs to be formally adopted and transposed. Watch for a formal Council vote and member-state implementation guidance in the coming weeks.

3. US Senate Consumer Data Protection Bill (Moran) — The bipartisan data privacy bill cited in TechPolicy.Press's March 2026 roundup could move to a committee markup hearing. If it advances, it would be the most significant federal data privacy legislation in years.

4. Enterprise AI Act Readiness Deadline Pressure — With 78% of enterprises still unprepared for EU AI Act obligations according to Vision Compliance, expect a wave of compliance announcements, legal guidance publications, and potential enforcement warnings from EU authorities as the nearest application deadlines approach.