Daily AI news, sourcedOfficial news, first rumor reports and substantive progress. Each entry keeps the information published at that time.
- OpenAI says METR and Redwood Research piloted an embedded eval of its internal monitoring
OpenAI says it invited METR and Redwood Research this summer to pilot an embedded evaluation of its internal monitoring system, giving outside researchers deeper access than typical external evals.
- OpenAI and Poland’s NASK partner on AI safety and security research
OpenAI and Poland’s NASK are partnering on AI safety, security and technical research to strengthen safeguards and support a more resilient digital ecosystem.
- OpenAI makes the Decisions API generally available for fast bounded choices with GPT-6 Luna
OpenAI made the Decisions API generally available. Built with GPT-6 Luna, it focuses the model on questions with bounded answers so apps can classify, route or score from text, images and structured inputs, which OpenAI says can be up to 10× faster than Luna through the Responses API. Since beta it added GPT-Live integration, prompt caching, US/EU residency, native JSON inputs, images in instructions and up to 400 questions per request. Input is $0.10 per million tokens with no output-token charge.
- OpenAI lets developers choose which API service tiers each project can use
OpenAI's API changelog for Oct 21 says developers can now choose which service tiers each project can use in Platform settings, and organization settings can restrict service tiers for all projects.
- OpenAI says GPT-Rosalind surfaced a drug-discovery lead linking cholesterol transport to neurodegeneration
OpenAI says GPT-Rosalind suggested that restoring cholesterol transport could protect neurons from degeneration. In mouse-brain data, disrupting DCTN1, a gene linked to the rare neurodegenerative disorder Perry syndrome, increased expression of genes involved in making cholesterol.
- OpenAI models are coming to Google Cloud, with GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna in private preview
OpenAI and Google Cloud are bringing OpenAI frontier models and Codex workflows to Gemini's Model Garden, available through Google Cloud Marketplace. According to The Decoder, GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna launch as a private preview for enterprise customers, making Google Cloud the first major cloud to offer Gemini, Claude and OpenAI models together.
- OpenAI says its Patch the Planet coalition coordinated fixes for two critical PHP flaws, one found by Calif using GPT-5.6 Sol
OpenAI says it spent two months coordinating maintainers, researchers and security providers on two critical vulnerabilities in PHP. Calif found one using GPT-5.6 Sol under OpenAI's Patch the Planet initiative, and Wiz independently found and reported both. PHP released fixes on Oct 21, and OpenAI is sharing a PHP version checker as a Codex skill.
- OpenAI explains how it holds distribution partners to consistent safeguards under its Usage Policies
OpenAI describes how it works with distribution partners to keep safeguards consistent, support responsible access to its models and clarify what its Usage Policies expect.
- OpenAI publishes part two of its democratic governance of frontier AI series, a blueprint for empowering cyber defenders
OpenAI outlines a blueprint for empowering cyber defenders with frontier AI, alongside stronger safeguards, modernized cybersecurity and more resilient public institutions.
- OpenAI releases MultiLangBench and says its latest models are far better at COBOL, Fortran and other low-resource languages
OpenAI introduced two evaluations for low-resource programming languages: MultiLangBench, which asks a model to build a SQLite-compatible database engine in each of 20 languages, and a COBOL migration test that requires byte-for-byte output parity after porting programs to C++, Java, Kotlin, Python or Rust. OpenAI says GPT-6 Astra averaged an 87.4% test pass rate on MultiLangBench, up from 26.8% for GPT-5.6 Sol, and GPT-6.1 Sol reached 86.6% on COBOL migration at low reasoning effort and 97.9% at higher effort, versus 5.4% for GPT-5.6 Sol. It is releasing the MultiLangBench setup and test suite on GitHub. The results are OpenAI's own.
- OpenAI's desktop app now lets you edit previous messages within a thread, Tibo says
Tibo (@thsottiaux) said on Day 14 of his daily updates that users can now edit previous messages within a thread in the desktop app, calling it a small quality-of-life improvement that had been requested for a long time and took too long to ship.
- OpenAI raises GPT-6.1 Sol default context to 500k in Codex and Work, and drops long-context fees on subscriptions
As OpenAI Codex head Tibo Sottiaux wrote on X (Day 7 of the shipping series), GPT-6.1 Sol's default context length in Codex and Work is now 500k tokens, up from 272k — about an 80% increase — to cut compaction and help long-context tasks. A follow-up post said OpenAI is not charging long-context fees for GPT-6.1 Sol when usage is drawn from a subscription. He said defaults will go higher later.
- OpenAI says it disrupted two AI-enabled false-front influence operations using fake journalists and a think tank
OpenAI reports disrupting two AI-enabled influence operations that used false-front journalists and a think tank to spread geopolitical messaging.
- OpenAI's Decisions API is now generally available, powered by GPT-6 Luna
OpenAI's developer account announced the Decisions API is here, following its September preview. The API is powered by GPT-6 Luna, accepts text and image inputs, and supports three kinds of outputs, starting with Predicates that estimate the probability that a statement holds. OpenAI engineer Charlie Guo celebrated the launch, noting one of his favorite demos missed the cut.
- OpenAI commits to a 28-day sprint: one visible Codex and ChatGPT Work improvement per day
OpenAI product and platform lead Thibault Sottiaux announced that for the next 28 days, Codex and ChatGPT Work will ship either one improvement that is clearly useful to most users or a full credit reset each day, per his announcement on October 5. The day before he had narrowed the focus to simplifying the product, raising limits, breaking through features and new models.
- Codex lets you edit the model and reasoning effort of scheduled tasks
OpenAI's developer account confirmed that the model and reasoning effort of Codex scheduled tasks can now be edited directly from the Edit task view. The update was highlighted in a post reshared by the official account.
- Codex Pro $200 subscriptions re-open to new subscribers tomorrow
OpenAI Codex lead Tibo Sottiaux said on X that Pro $200 subscriptions will re-open to new subscribers tomorrow.
- Codex 0.159.0 adds instant interrupt and refreshed session welcome
OpenAI's Codex rust-v0.159.0 adds opt-in instant_interrupt so new input can steer the model mid-response or during long code-mode calls, and gives new sessions a compact welcome screen, per the GitHub release notes.
- OpenAI lays out safety-case approach for frontier AI training
OpenAI published 'Towards safety cases for frontier AI training', outlining how it thinks about securing frontier RL training runs, per its official X account.
- OpenAI details textGrain watermarking for EU AI Act, detector access starts with researchers
OpenAI published its approach to text watermarking under the EU AI Act, built on textGrain — an invisible statistical signal in the model's word choices. Starting immediately, API customers globally can opt in for select models (off by default), eligible ChatGPT and Codex output in the EU gets the watermark over the coming weeks, and detector access opens to approved researchers and expert organizations only. At a 1% target false-positive rate, detection worked on about 80% of 200-token passages and 95% of 400-token passages, but replacing 25% of words cut detection to 17%. OpenAI says textGrain matched or exceeded SynthID for text in its tests and plans to open-source the technology.