HN Brief - 2026-02-28
HN Brief - 2026-02-28
Welcome to the daily Hacker News brief! Here are the top stories and discussions from today.
AI & Technology Policy
We Will Not Be Divided (1560 points)
Summary: OpenAI and Google employees released a collective statement refusing to cooperate with the US Department of War’s demands to deploy AI models for domestic mass surveillance and fully autonomous killing systems without human oversight. The protests come amid escalating tensions between AI companies and the government over ethical use of AI in military applications.
Comments: Highly polarized debate across 500+ comments. Supporters praise the engineers for taking a moral stand against military AI applications. Critics argue this is hypocrisy from companies building these technologies while claiming ethical concerns. Many discuss broader implications for US economy and rule of law, worrying about government procurement rules being abused to punish companies for perceived lack of loyalty. Some point to Five Eyes agreements that render domestic surveillance bans ineffective since countries can spy on each other’s citizens and share data.
OpenAI agrees with Dept. of War to deploy models in their classified network (599 points)
Summary: Sam Altman announces OpenAI partnership with US Department of War for classified network deployment, contrasting with Anthropic’s refusal to participate under similar terms.
Comments: Sharp criticism from users who signed the “We Will Not Be Divided” letter. Many question how OpenAI employees can continue working there after this deal. Some point to an Under Secretary’s explanation that OpenAI’s agreement references existing legal authorities while Anthropic’s involved purely subjective conditions. Users are cancelling subscriptions in protest.
A new California law says all operating systems need to have age verification (570 points)
Summary: California passed legislation requiring all operating systems (including Linux) to include age verification/attestation features at account setup for users.
Comments: Massive criticism of the law’s technical feasibility. Users point out absurd scenarios: embedded systems with no UI, servers in data centers, smart appliances, cars. Developers create humorous “age ratings” for Linux commands (rm - all ages, cat - 18+ for displaying “porn”, sudo - 18+). Many argue it’s poorly drafted and technically impossible to implement, suggesting companies will simply add “not for use in California” disclaimers.
OpenAI raises $110B on $730B pre-money valuation (483 points)
Summary: OpenAI raised $110 billion at a $730 billion pre-money valuation from SoftBank ($30B), NVIDIA ($30B), and Amazon ($50B). Microsoft notably did not invest.
Comments: Heavy skepticism about valuation (~66x revenue vs Microsoft’s 12x). Users call it a “circular investment” - money committed to AWS and NVIDIA hardware that flows back to them. Many compare to WeWork’s doomed IPO, questioning if OpenAI has a sustainable moat without network effects.
Security & Privacy
Don’t use passkeys for encrypting user data (141 points)
Summary: Technical blog post arguing against using passkeys as encryption keys for user data, highlighting risks of data loss when users lose access to passkeys.
Comments: Debate split. Some agree passkeys have too many footguns - users accidentally create them in wrong places (iOS webview vs Chrome vs password manager) and can’t recover them. Others argue passkeys are actually a step forward because users regularly lose encryption keys anyway, and automatic syncing through password managers or cloud providers reduces data loss in practice. Author criticized for writing generic anti-encryption advice rather than passkey-specific concerns.
Geopolitics & War
U.S. and Israel Conduct Strikes on Iran (226 points)
Summary: Major military action involving US and Israel strikes on Iran, with ongoing developments.
Comments: Limited comments in preview; story appears to be actively developing with multiple perspectives being discussed.
Tech Tools & Projects
How do I cancel my ChatGPT subscription? (723 points)
Summary: Ironical link to OpenAI’s subscription cancellation help page, posted in response to OpenAI’s Department of War deal.
Comments: Practical discussion on cancellation process and data export. Users share experiences: some cancelled earlier due to poor value, others cancelling now as moral protest. Discussion of moving to Claude, Grok, or local models. Stories of successful chargebacks when OpenAI kept charging after cancellation. Recommendation to use local models (Qwen3.5 27B/35B on Mac) to avoid corporate AI dependence.
Show HN: Manim-Web (99 points)
Summary: Browser-based port of 3Blue1Brown’s Manim math animation engine, running entirely client-side in browser at 60fps using Canvas/WebGL. Includes LaTeX rendering via MathJax/KaTeX.
Comments: Enthusiastic reception. Users excited about zero-setup alternative to notoriously painful Python installation (Python, FFmpeg, Cairo, full LaTeX). Discussion of interactive possibilities - objects draggable/clickable, embeddable in React/Vue apps. Interest in py2ts converter for migrating existing scripts.
Web & Infrastructure
A better streams API is possible for JavaScript (406 points)
Summary: Cloudflare’s proposal for improved Web Streams API to make it more ergonomic and powerful for JavaScript developers.
Comments: Technical discussion about current API limitations and proposed solutions. Interest in better abstractions for streaming data handling.
History & Science
Croatia declared free of landmines after 31 years (234 points)
Summary: Croatia officially declared mine-free after 31 years of demining efforts since the 1990s war following independence and subsequent conflicts.
Comments: Celebration of this achievement. Discussion of the scale and difficulty of the task. Reflection on the persistence of wartime dangers long after conflicts end. Some sharing personal connections to the region and memories of the war years.
NASA announces overhaul of Artemis program (253 points)
Summary: NASA restructuring Artemis program, moving to more frequent SLS launches (every 10 months). Artemis III now becomes an Earth-orbit docking test before lunar landing attempts. Multiple landing attempts planned for 2028.
Comments: Comparison with SpaceX’s approach - Starship launches 11 times in 2 years (many blowing up) vs Artemis trying for near-perfection. SpaceX <$10B vs Artemis $92B raises questions about capital/time efficiency. Praise for increased launch frequency improving reliability through learning. Concern for astronauts on upcoming trans-lunar flight given Boeing’s issues and political climate. Respect for Apollo engineers expressed.
Summary by Category
- AI & Tech Policy: 4 stories - dominated by ethical AI use debates and regulatory concerns
- Security & Privacy: 1 story - passkeys for encryption
- Geopolitics & War: 1 story - escalating tensions and military AI debates
- Tech Tools & Projects: 2 stories - recovery tools, versioning, math animations
- Web & Infrastructure: 1 story - JavaScript streams API
- History & Science: 2 stories - Croatia demining, NASA Artemis
Key Themes:
- Massive backlash against OpenAI’s Department of War deal
- Support for Anthropic’s principled stance
- California’s age verification law widely criticized as technically impossible
- Skepticism about OpenAI’s $730B valuation
Want more HN briefs? Come back tomorrow for the next roundup!