The United States Congress moved closer to passing its first major federal AI regulations this week. A key proposal advanced by the House Energy and Commerce Committee would block states from creating their own AI laws for 10 years. Supporters say this moratorium prevents a confusing patchwork of state rules, making it easier for companies to comply. But opponents argue it could delay important protections if Congress fails to act quickly.

Under the new Trump administration Executive Order, AI developers no longer need to conduct federal red-teaming (security testing) or submit bias audit reports unless required by specific agencies like the FDA. The order reverses Biden-era rules that required watermarking AI-generated content and sharing safety data. Tech firms praised the changes for reducing costs, while consumer groups warned they might lead to riskier AI deployments.

Congress passed the TAKE IT DOWN Act, which targets AI deepfakes used to create fake nude images. Platforms must now remove such content within 48 hours of reporting. Though aimed at protecting privacy, digital rights groups say the law’s vague language could let platforms censor legitimate posts, especially those criticizing public figures.

Debates continued about how to regulate agentic AI (self-acting systems). Some lawmakers pushed for risk-based frameworks, where stricter rules apply only to high-harm uses like healthcare or hiring. Others favored light-touch regulation to keep the U.S. competitive with China and the EU in AI development.

Meanwhile, think tanks urged policymakers to avoid copying the EU’s strict AI Act, arguing it might stifle American innovation. Instead, they recommended focusing on sector-specific guidelines (e.g., finance or transportation) rather than blanket AI laws.

Globally, no other countries announced major AI policy changes this week. Analysts note that the U.S. debates could influence international standards, as many nations are still shaping their own AI rules.