According to @gdb, OpenAI teased “fly with Codex” in a video, hinting at code-to-control demos. As reported by OpenAI’s post, developers await details.

According to OpenAINewsroom, OpenAI released a frontier AI safety governance blueprint to guide US institutions after a new cyber EO.

According to @godofprompt, Leni targets finance with verified outputs and rigorous reasoning to cut costly errors in institutional investment workflows.

According to @godofprompt, Zeely will launch emotion-adaptive AI in two days, reading tone and adapting responses, with 2x credits on launch day.

According to emollick, PNAS ranked a study on persuading LLMs to comply with harmful requests, highlighting jailbreak risks across top models.

According to TheRundownAI, Microsoft revealed new AI models and agents at Build, Trump eased AI review rules, and Claude Design added slide decks.

According to OpenAINewsroom, Codex tops 5M weekly users, expanding from coding to research, analysis, content, and ops productivity.

According to @gdb, OpenAI is rolling out Codex Sites to Business and Enterprise to turn team content into shareable web apps, per OpenAI on X.

According to TheRundownAI, OpenAI’s top customer uses 100B tokens monthly, and Sam Altman says it is not the global leader, indicating massive demand.

According to @OpenAI, Codex now installs role-based plugins with no code, tapping 62 apps and 110 skills for sales, analytics, design, and investing.

According to TheRundownAI, OpenAI’s Codex adds Sites, 6 role plugins, and Annotations to boost no-code workflows for 5M users, 20% non-developers.

According to OpenAI... Frontier models and Codex hit AWS via Bedrock, enabling governed enterprise deployments and future Daybreak cybersecurity on AWS.

According to @OpenAI, frontier models and Codex are now on AWS Bedrock, with Daybreak cybersecurity capabilities planned next.

According to @sama, OpenAI Foundation launched $130M grants for bio and cyber resilience, model safety, and youth impact, accelerating AI risk management.

According to God of Prompt, HydraDB raised $6.5M to power graph-native memory for agents, tackling stateless production failures and enabling precise context.

According to emollick, walls of text hinder AI UX by raising cognitive load and reducing engagement, prompting concise, structured prompt design.

According to TheRundownAI, Higgsfield and Claude enable a DIY video AI workstation, new tools emerge, and ex DeepMind vets build self improving science AI.

According to sama, OpenAI launched Rosalind to harden biodefense with guardrails, evaluations, and partnerships for safer bio research.

According to @gdb, GPT Realtime 2 enables full voice computer control, showcasing low latency, multimodal agents with OS actions in a live demo.

According to @emollick, OpenAI and Anthropic are accelerating releases with 3+ point jumps on the Artificial Analysis index, indicating faster capability gains.