AI News

What an Independent Frontier-AI Assessor Is Actually Being Asked to Do An independent frontier-AI assessment is a structured inquiry into specific safety claims, safeguards, evaluation results, incident evidence, or risk-management arguments made by an AI developer or deployer. In OpenAI’s September 2026 proposal on third-party assessments, the core object of…

What a Voice-to-Work Workflow Is—and What It Is Not A Voice-to-Work workflow is a governed sequence for using spoken instructions to start useful work in ChatGPT, retrieve information from authorized apps, create drafts or artifacts, review proposed actions, approve only the intended external steps, capture evidence of what happened, and…

Prompt-cache FinOps is a governance system, not a hit-rate contest Prompt caching for GPT-6 changes the unit economics of repeated input, but the operational problem is larger than making a dashboard line go up. In this playbook, prompt-cache FinOps means the combined discipline of cost attribution, tenant accounting, privacy governance,…

Start with an experiment, not a model swap GPT-6 Sol and GPT-6 Luna should enter an existing production workflow as candidates in an evidence-led rollout experiment, not as a direct replacement for the current model ID. OpenAI announced Sol and Luna on September 22, 2026 as faster, lower-cost members of…

Why AI-generated mathematics claims need a slower review lane OpenAI’s September 21, 2026 announcement about an Advisory Group on Mathematics and Artificial Intelligence created a high-stakes review problem for mathematicians, AI labs, universities, funders, journals, enterprise research teams, and anyone tempted to treat model-generated mathematics as immediately settled knowledge. OpenAI…

What “source-linked institutional memory” means in practice Source-linked institutional memory is a structured memory layer that lets ChatGPT, Codex, or an API-based assistant answer organizational questions by tracing each important claim back to approved source material. In a serious implementation, “memory” is not a vague bucket of past chats or…

Why Codex 0.155.0 task lifecycle controls need a defensive playbook Codex CLI 0.155.0, listed in OpenAI’s official ChatGPT and Codex changelog for September 17, 2026, adds several features that change how teams should manage local task state: task hiding, archiving, deletion, worktree ownership details, confirmed deletion of clean managed worktrees,…

What Codex CLI 0.155 experimental /voice is for Codex CLI 0.155.0 added an experimental /voice conversation mode for supported builds, according to OpenAI’s official ChatGPT and Codex changelog and the Codex 0.155.0 release notes. The practical idea is simple: instead of typing every instruction into the local terminal user interface,…

Why this playbook starts with governance, not campaign optimization OpenAI’s September 16 advertising announcements create a new operating surface for marketers, growth teams, agencies, ecommerce teams, CRM owners, and compliance reviewers: ads can appear in ChatGPT, advertisers can manage campaigns through natural-language requests, AI can suggest creative assets, selected advertisers…

What the ChatGPT Work cloud browser is—and why signed-in tasks need a different operating model OpenAI describes the ChatGPT Work cloud browser as a separate remote browser that ChatGPT can use to work on supported websites on your behalf. The important word is “separate”: it is not your laptop browser,…