Trump Plan Pushes AI Growth by Slashing Federal Rules

A study in Consciousness and Cognition finds participants remember words better when they draw them instead of writing them. Across repeated tasks, sketchers consistently beat writers on recall tests.

Researchers used multidimensional experience sampling to examine thought quality and focus. They observed drawing spark richer visual, motor, and spatial processing, while writing remained linear and verbal.

The article argues that large language models bypass the messy, generative phase of creation, leaving users to merely tweak finished text. When AI handles the making, the author warns the deeper cognitive work that cements knowledge stays with the machine, not the mind.

The Trump administration released a 28-page “Winning the Race: America’s AI Action Plan” laying out more than 90 steps to strip away rules that slow AI deployment. The document relies on executive orders and agency mandates to clear what it calls regulatory roadblocks.

Key actions fast-track permits for data centers and chip fabs, promote exports of “secure, full-stack AI packages” to allies, and remove DEI language from federal AI contracts. The plan positions Commerce and State to steer overseas adoption of U.S. technology while sidestepping foreign rules.

Opponents warn the rush sacrifices environmental safeguards, worker retraining, and bias protections as data centers already consume 176 TWh of U.S. electricity and emit 105 million tonnes of CO₂. Civil-rights groups add that the blueprint could override state laws and cut funding to jurisdictions that impose stronger AI protections.

A small group of U.S. judges are using generative AI to summarize cases, draft routine orders, and prepare hearing questions. Early experiments promise faster legal research for chronically backlogged courts.

Yet judges in New Jersey, Mississippi, and Georgia have already issued opinions containing AI-fabricated citations and factual mistakes, then quietly retracted or revised them. Lawyers get sanctioned for similar missteps, but judges face minimal accountability and rarely disclose how the errors occurred.

Researchers and judges admit the boundary between “safe” administrative tasks and judgment-heavy decisions is blurry, and February guidelines from the Sedona Conference still flag hallucinations as unresolved. Observers warn that judicial AI mistakes can immediately shape the law and erode public trust far more than attorney blunders.

Japanese eCommerce group Rakuten is embedding its new agentic AI platform across eCommerce, FinTech, telecom, travel and more, with a launch on Rakuten Ichiba slated for the fall.

The system lets shoppers snap a photo of an item, then surfaces similar products in their price range and guides them straight to purchase pages while learning from transaction history.

Rakuten built its own large language model optimized for Japanese language, enabling the platform to act as a “brain twin” that learns preferences and unifies fragmented data in a single interface.

Addressing Black Hat USA 2025, former New York Times reporter Nicole Perlroth declared that AI has pushed cyber threats beyond the reach of conventional defenses. She said malware now operates autonomously, ransomware functions like a subscription, and AI is distorting reality.

Perlroth recounted a decade of escalating attacks—from the Times newsroom hack to hospital-crippling ransomware—to show adversaries learning from each incident. She added that hackers already deploy AI to map critical assets, write persuasive phishing lures, and run chatbots that hard-ball ransom negotiations.

She warned defenders are falling behind as AI models like Claude win hacking contests and agentic workflows become new targets. Perlroth said the chance to “steer AI” is shrinking fast and that courage, not code alone, will decide the outcome.