TypeSafe's Jev, a "System One" decision model that returns likelihoods over a list of options in about 270ms per decision and roughly half a cent per Manic Miner cavern attempt, was tested by a develo…
A developer's attempt to train a reinforcement-learning agent to complete the 1983 ZX Spectrum game Manic Miner failed, first with PPO over joystick inputs and then with move macros, according to a fi…
Google's Gemini model flagged a deceptive YouTube ad as a policy violation in seconds, yet Google's human review process approved the same ad twice after multiple user reports, according to a first-pe…
Developer atomic14 released an updated AI-text detector at meatbag.atomic14.com that can now identify both Claude Opus 5 and ChatGPT output, achieving 93.5% accuracy on held-out authors for ChatGPT an…
On 2 August 2026, Anthropic began watermarking text produced by new Claude models, and a developer known as atomic14 has built a web-based detector that can identify text written by Claude Opus 5 with…
Developer atomic14, author of the 'harmless' project on GitHub, reports frustration with Anthropic's Claude AI assistant after 700 commits, citing that Claude ignored instructions to use ASD-STE100 Si…
A new paper, 'The AI Layoff Trap,' argues that businesses face an automation arms race that will eventually destroy their customer base, and proposes a Pigouvian tax on automation. However, the author…
A developer reports that changing Claude's output style to Simplified Technical English (ASD-STE100) does not improve the quality of its prose, which remains 'vacuous and empty' despite producing good…
Chris Thompson, a software engineer, used Anthropic's Claude AI to rebuild his 2006 travel blog as a static site at travellingchris.atomic14.com, recovering 766 of 778 images from 182 posts across 14 …
A developer asked AI to design 12 ESP32 dev boards, but the AI suggested writing deterministic scripts instead, which the developer did. The resulting open-source project, atomic14/kicad-esp32-dev-boa…