GoKawiil - Tech News

1.

Show HN: misa77 - a codec that decodes 2x faster than LZ4 (at better ratios) (news.ycombinator.com)

2026-07-15 | tags: compression, mb mb, mb silesia

2.

Unauthorized alert sent to cell phones across Brazil (news.ycombinator.com)

2026-06-20 | by Mariana Catacci | get Smartphone Security Alert System → | tags: national civil defense, misantropi4, wireless emergency alerts

3.

Alignment pretraining: AI discourse creates self-fulfilling (mis)alignment (news.ycombinator.com)

2026-05-18 | get AI Alignment Book → | tags: alignment pretraining, ai discourse, self-fulfilling

4.

An idiot’s guide to lead optimisation for proteins (news.ycombinator.com)

2026-05-11 | get Protein Structure Model Kit → | tags: lead optimisation, cradles-1, machine learning

5.

Anthropic says ‘evil’ portrayals of AI were responsible for Claude’s blackmail attempts (techcrunch.com)

2026-05-10 | by Anthony Ha | get AI Ethics and Safety Book → | tags: anthropic, claude, ai models

6.

Teaching Claude Why (news.ycombinator.com)

2026-05-08 | get AI Chatbot Development Kit → | tags: claude, ai models, agentic misalignment

7.

Continuously graded-doped SnO<sub>2</sub> for efficient n–i–p perovskite solar cells (feeds.nature.com)

2026-04-30 | by Wang | get Perovskite Solar Cell Kit → | tags: perovskite solar cells, sno2, electron transport layer

8.

Spotify is giving artists a layer of protection against misattributed songs (feeds.feedburner.com)

2026-03-24 | get Spotify Artist Protection Tool → | tags: spotify, blaze foley, artist profile

9.

Two studies in compiler optimisations (news.ycombinator.com)

2026-03-22 | get Compiler Optimization Guidebook → | tags: llvm, gcc, clang

10.

Poking holes into bytecode with peephole optimisations (news.ycombinator.com)

2026-01-15 | get Peephole Optimizations → | tags: optimisation, peephole, registerallocator

11.

Training large language models on narrow tasks can lead to broad misalignment (feeds.nature.com)

2026-01-14 | by Betley | get Language Model → | tags: emergent, emergent misalignment, finetuning

12.

Show HN: Misata – synthetic data engine using LLM and Vectorized NumPy (news.ycombinator.com)

2025-12-16 | get Misata → | tags: generate, generate story, misata

13.

Nothing reportedly halts Nothing OS 4.0 rollout for ‘urgent fix’ (Update: Statement) (androidauthority.com)

2025-12-11 | get Nothing OS → | tags: android, apps, authority

14.

Advent of Compiler Optimisations 2025 (news.ycombinator.com)

2025-12-02 | get Compiler Optimizations → | tags: advent, blog, compiler

15.

Anthropic Researchers Startled When an AI Model Turned Evil and Told a User to Drink Bleach (futurism.com)

2025-11-29 | by Sharon Adarlo | get AI model → | tags: ai model, anthropic, human

16.

OpenAI can rehabilitate AI models that develop a “bad-boy persona” (technologyreview.com)

2025-10-31 | by Peter Hall | get The Power of Habit book → | tags: bad, fine, misalignment

17.

Agentic Misalignment: How LLMs could be insider threats (news.ycombinator.com)

2025-10-31 | get Google Nest Hub → | tags: agentic, goal, misalignment

18.

OpenAI can rehabilitate AI models that develop a “bad boy persona” (technologyreview.com)

2025-10-31 | by Peter Hall | get The Power of Habit book → | tags: bad, fine, misalignment