New Paper Compares OpenAI, Anthropic and Alibaba Model Specs via Expert Interviews
S_OhEigeartaigh · x · 2026-11-13
A paper led by Edward Kemberley, with contributions from SOhEigeartaigh, examines the role of Model Specs in AI development through a comparative analysis of the OpenAI, Anthropic, and Alibaba specs plus expert interviews.
Core argument: as AI systems grow more capable and autonomous, behavioral choices — what a model should do, refuse, and whose instructions to follow — become more consequential, and model specs make those choices explicit.
More from Safety
- Exa rebuilds its AI search engine inside the EU, storing and processing in Stockholm — kimmonismus · 2026-11-13
- Microsoft Research: editing fewer than 10 dormant neurons can backdoor an LLM agent — DimitrisPapail · 2026-11-13
- DeepMind: agent-swarm attacks trick best-defended Gemini model on 34% of tasks — rohanpaul_ai · 2026-11-13
- Prediction: mandatory interception for inference, telemetry for GPUs — l4rz · 2026-11-13
- Lost updates in four MCP servers: GitHub, Grafana, PostHog and Kubernetes — Miserable_Bed_219 · 2026-11-13
- A 'verified' email can still be from a stranger: DMARC alignment for agents — Spacesh1psoda · 2026-11-13