Newsletter Archive

Pope Leo XIV issued 'Magnifica Humanitas,' his first encyclical, calling AI an "instrument of domination, exclusion and death" and demanding robust regulation, prohibition of autonomous lethal weapons, and developer accountability to common good over profit. Anthropic and other tech leaders attended the Vatican presentation.

Fortune · 6 min read Policy

Alphabet is monetizing AI through hardware (TPUs), joint ventures, and new Gemini pricing tiers—expanding beyond cloud services to compete with NVIDIA and capture more of the AI infrastructure market.

Simply Wall St · 7 min read Industry

China is accelerating development of homegrown AI chips and large language models to reduce dependence on Western technology amid export restrictions. The strategy spans semiconductor design, foundry partnerships, and LLM development to create a complete domestic AI stack.

Digitimes · 3 min read Hardware

Researchers measured that frontier reasoning models waste 61–93% of their chain-of-thought steps on redundant thinking—with most answers determinable from a single reasoning segment. The finding reveals a structural inefficiency in how current models use length-based rewards, suggesting massive latency and compute savings are possible through smarter early stopping or model redesign.

arXiv AI · 4 min (abstract) Research

People who phrase requests with clear logic and structured thinking get dramatically better ChatGPT results. The piece explores why clarity of thought patterns matter more than fancy prompting tricks and shows specific techniques to apply.

Tom's Guide · 1 min read Tools

OmniVoice Studio bundles voice cloning, video dubbing, real-time dictation, and speaker diarization entirely on local hardware—no cloud, no subscriptions, no API keys. Supports 646 languages, integrates with Claude via MCP, and built on diffusion-based TTS using zero-shot learning from 3-second voice samples.

MarkTechPost · 9 min read Tools

China is now barring top AI researchers at private companies like Alibaba and DeepSeek from traveling overseas, escalating tech talent controls as Beijing tries to prevent brain drain and accelerate its AI capabilities against US competition.

Bloomberg Tech · 1 min read Policy

CISA has lost roughly a third of its workforce since early 2025 and faces a diminished role just as the U.S. government scrambles to defend critical infrastructure against AI-powered cyberattacks. Former officials and industry leaders warn the weakened agency can no longer adequately support utilities, banks, and other critical sectors preparing for a wave of AI-fueled threats.

Axios · 5 min read Policy