- OpenAI: Introducing GPT-5.4 mini and nano (Mar. 16, 2026)
Mini nears GPT‑5.4 performance for complex tasks, and nano minimizes latency and cost for simple, high-volume workloads. - Google: Gemini 3.1 Flash Lite: Our most cost-effective AI model yet (Mar. 3, 2026)
Gemini 3.1 Flash-Lite is a fast, low-cost model previewing in AI Studio and Vertex AI, at $0.25/1M input, $1.50/1M output. It suits high-volume tasks like translation, moderation, and UI. - Anthropic: Labor market impacts of AI: A new measure and early evidence (Mar. 5, 2026)
The paper introduces observed exposure, a measure combining LLM capability and real-world usage, weighting automated, work-related tasks. - Tyler Cowen: The AI arms race (Mar. 17, 2026)
The biggest risk is a government with the most powerful AI becoming the bad guy, not AI companies. - Inside Higher Ed: Journal Submissions Riddled With AI-Created Fake Citations (Mar. 6, 2026)
Academic journals are seeing more AI-generated, fake citations that look convincing, cite nonexistent sources, and misattribute work. - Peter Lavigne: Toward automated verification of unreviewed AI-generated code (Mar. 16, 2026)
An experiment shows unreviewed AI-generated code can be trusted if automatically verified via property-based tests, mutation testing, side-effect constraints, type-checking, and linting. - Money Control: How Palantir and Anthropic AI helped the US hit 1,000 Iran targets in 24 hours (Mar. 6, 2026)
The US military used Palantir’s Maven, with Anthropic’s Claude AI, to generate, prioritise, and assess about 1,000 strike targets against Iran in the first day. The Pentagon is phasing Anthropic out, but may retain use during transition. - NY Times Opinion: A.I. Is Coming for Politics (Mar. 17, 2026)
A.I., led by major tech firms, risks concentrating economic, political, and cultural power, undermining democracy and increasing inequality. It can manipulate voters, deanonymize people, and centralize control. - NY Times Opinion: The A.I. Labor Crisis Is Coming. This Is the Solution. (Mar. 6, 2026)
A.I. will displace millions of workers unless America builds a public-private transition system, employer-led skills data, and fast, job-linked training. - Reuters: xAI loses bid to halt California AI data disclosure law (Mar. 5, 2026)
A federal judge denied xAI’s bid to block California’s law requiring public summaries of AI training data. - Alex Tabarrok: Claude on NY’s Senate Bill S7263 (Mar. 5, 2026)
A proposed New York bill would make it a crime for chatbots to give substantive advice that would be unauthorized practice by a person. Its vague test risks overbroad enforcement, chilling useful AI help, and harming low-income, rural users. - WSJ: Can Nvidia’s Dominance Survive the Sea Change Under Way in AI Computing? (Mar. 16, 2026)
Nvidia is shifting GTC’s focus from GPUs to inference, as AI agents drive revenue and need efficient, memory-rich hardware.