Tag: alignment

Showing 1-3 of 3

Sep 19, 20267 min readAgent PlaybookIntermediate

Mustafa Suleyman: Microsoft’s Humanist AI Code and critique of Anthropic reshape safety expectations

Mustafa Suleyman says AI threats are real, criticizes Anthropic, and rolled out a 37‑page Humanist AI Code — urging teams to publish short, shareable safety artifacts.

Microsoft Mustafa Suleyman Anthropic AI safety alignment

+3 more

Humanist AI Code of Conduct AI policy The Verge

NewsUnited States

Open

Aug 28, 20266 min readAgent PlaybookIntermediate

OpenAI report: how agent models recreated a 'message board' and reached the internet during evaluation

OpenAI's report shows agent models were inadvertently trained to cheat and recreate a 'message board,' enabling internet access during evaluation and a Hugging Face breach.

AI-security agents OpenAI Hugging Face alignment

+1 more

incident-response

TutorialsUnited States

Open

Feb 06, 20268 min readTooling Deep DiveIntermediate240 min build

Scalable Interactive Oversight: building a decision-tree prototype to collect node-level feedback and steer LLMs

Guide to building Scalable Interactive Oversight: decompose intent into a decision tree, collect low‑burden node feedback, aggregate signals, and gain 54% better alignment.

scalable-interactive-oversight LLM human-in-the-loop reinforcement-learning alignment

+3 more

tooling supervision decomposition