- AI Week in Review
- Posts
- AI Week in Review - 9/20/25
AI Week in Review - 9/20/25
Public Sector
Chris Kraft
September 20, 2025
The Details
Federal
The U.S. and U.K. signed a landmark deal to deepen cooperation on AI, quantum, and nuclear technologies, promoting shared research, standards, and energy security.
The FTC has launched an inquiry into AI companion chatbots, seeking information from seven firms about how they protect children, limit harms, and inform users/parents.
The U.S. Air Force’s DASH‑2 experiment proved AI can accelerate battle‑management decisions—producing 30× more solution options in under ten seconds while matching human judgment in accuracy.
Acting Federal CISO Mike Duffy says U.S. agencies are accelerating AI adoption for cyber defense—boosting vulnerability detection, streamlining toolsets, and updating policies like OMB Circular A‑130 to match modern threat landscapes.
The CDC says its rollout of GenAI (including ChatGPT) to its ~10,000 workers has saved about 41,000 work hours and achieved over 500% return on investment.
Senator Mark Kelly’s AI for America roadmap proposes an “AI Horizon Fund” financed by AI companies to support worker retraining, infrastructure demands (e.g. energy), responsible deployment, and ensure AI benefits reach all Americans.
State / Local
Seattle’s 2025–2026 AI Plan outlines a people-centered, equity-driven strategy to responsibly scale AI across city services—focusing on public safety, housing, accessibility, workforce upskilling, and strong governance.
The Glassboro, NJ school district is the first in the U.S. to deploy an AI surveillance system across hundreds of cameras to detect firearms and automatically alert staff, law enforcement, and lock down buildings.
Ohio launched Safeguard Ohio, an AI-powered app letting people anonymously report suspicious activity with multimedia (video, photo, audio), which prompts follow-ups and routes tips to law enforcement.
Colorado is now 8th in the U.S. for per-capita AI use. Denver has ~8,100 AI specialists (ranked 14th in North America) and is emerging as a strong hub for AI talent.
International
Egypt leads African nations in AI readiness, followed by Mauritius and South Africa; smaller states like Seychelles also score high by combining governance, infrastructure, and policy alignment.
Portugal’s Ministry of State Reform will unveil its National Agenda for Artificial Intelligence in Q4 2025, including the natural language model “Amalia” in European Portuguese among proposed AI initiatives.
The UK’s MoD is testing its £1 billion Digital Targeting Web—an AI‑enabled system linking sensors, deciders, and effectors to speed decision‑making in cyber, electromagnetic, and physical domains.
New South Wales passed laws criminalizing the creation or sharing of sexually explicit deepfake content and non‑consensual intimate audio, with penalties up to three years, to protect women and girls.
Albania appointed “Diella,” an AI‑generated virtual assistant, as its first cabinet‑level minister (for public procurement), aiming to eliminate corruption in tenders and increase efficiency and transparency.
Japan will fund development of a fully domestic AI model—trained on Japanese data and hosted in-country—to reduce reliance on U.S./Chinese models and ensure cultural, security, and data sovereignty.
The UK MoD has struck a partnership with Palantir to invest up to £1.5 billion, creating ~350 jobs, to build AI‑powered defense tools—targeting, decision support, and data fusion—for faster threat response.
India will set up AI R&D hubs in “Institutes of Excellence” focusing on agriculture, education, health, urban living, space and nuclear sectors; ITIs will be upgraded into AI‑training centers.
OpenAI announced Stargate UK, a partnership with NVIDIA and Nscale to build sovereign AI compute infrastructure in the UK — hosting advanced models locally for regulated industries and public services.
Italy has passed a national AI law—the first in the EU—aligned with the AI Act that mandates oversight, criminal penalties for misuse (e.g. deepfakes), child protections, and €1B for innovation.
Everything Else
At Meta Connect 2025, Mark Zuckerberg’s live demos of the $799 Ray‑Ban Display glasses failed twice—once during a cooking assistant test, once answering a video call, both blamed on flaky WiFi.
Anthropic’s Sep 2025 Economic Index finds AI adoption growing rapidly but unevenly: wealthier countries and enterprise users lead in automation; consumer use still skewed toward coding and augmentation.
Elon Musk’s xAI raised $10 billion, boosting its valuation to $200 billion, as it readies to scale infrastructure and compete with top AI firms.