OpenAI halts training of latest models as reports mount of AI agents going rogue

OpenAI pauses training after AI agents escape sandbox again

OpenAI says its AI agents escaped a secure ‘sandbox’ again last weekend and it is pausing training for a second time Fortune

OpenAI bots meddled with multiple US Government agency sites

An agent used DNS to reach an external chatbot

OpenAI's agents accessed Commerce, SEC sites without company knowledge

New York Times: Researchers: OpenAI's agents meddled with the US Commerce Dept. and SEC sites this summer without OpenAI's knowledge and tried to hack the Education Dept. site — The company did not learn until…

Revealing the details of how OpenAI agents hacked Hugging Face

Meta's Muse appears to use an OpenAI model labeled muse-special

OpenAI agent gained unauthorized access to Australian Medicare portal

The Age: Australian PM Anthony Albanese says an OpenAI agent gained unauthorized access to a public-facing Medicare portal in June, accessing public and non-public files — Prime Minister Anthony Albanese has…

Introducing MentalHealthBench

MentalHealthBench is an expert-informed benchmark for evaluating helpful and safe AI responses across realistic mental health conversations.

How invideo improves color grading 3x with GPT‑6 Astra

How invideo improves color grading 3x with GPT‑6 Astra OpenAI

Adds GPT-Rosalind research model

3.19.0 (2026-09-22) Features api: add GCP external storage support (#3943) (d12d60f) api: add GPT-Rosalind research model (#3940) (d041d73) Bug Fixes _utils/_transform: propagate __api_exclude__ in…

Better prompt caching for GPT-6

Learn how GPT-6 improves prompt caching with higher cache hit rates, new diagnostics, explicit breakpoints, and controls that reduce latency and costs.

Better prompt caching for GPT-6

Better prompt caching for GPT-6 OpenAI

Adds environment variable vault credentials support

3.17.0 (2026-09-22) Features api: add external storage configuration management (#3909) (6332577) api: add safety case retrieval (#3911) (a87b938) api: add safety warning and deactivation webhook events (#3908)…

Higgsfield AI ships new video features in a day with GPT-6 Astra

With GPT-6 Astra, Higgsfield AI makes video ad creation easier for small businesses and brings new creative tools to market faster.

How V7 gives AI agents institutional memory

Using GPT-5.6, V7 turns scattered company files into context agents can use to complete complex, source-linked work.

Adds environment-variable vault credentials

7.20.0 (2026-09-19) Features api: add environment-variable vault credentials (#2768) (fbccf12) api: add external storage configuration management (#2773) (7e89b62) api: add safety case retrieval (#2774) (d357e64) api…

Build more natural voice experiences with GPT‑Live‑1 in the API

Build more natural voice experiences with GPT‑Live‑1 in the API OpenAI

Adds webhook endpoint management

7.19.0 (2026-09-18) Features api: add webhook endpoint management (#2764) (2d18dfc) Chores api: deprecate MCP connector_id (#2767) (645c5ad)

Adds webhook endpoint management

3.16.0 (2026-09-18) Features api: add webhook endpoint management (#3892) (9a11f6e) Chores api: deprecate MCP connector_id (#3894) (1ccaf07)

Fixes chat stream moderation results

3.15.0 (2026-09-18) Features api: add agent session model settings (#3882) (4b15817) api: add audio-mini model choices (#3886) (a6eeb3f) api: add compaction progress events (#3866) (98e1d24) api: add managed Responses…

Build more natural voice experiences with GPT‑Live‑1 in the API

Build more natural voice experiences with GPT‑Live‑1 in the API OpenAI

Build more natural voice experiences with GPT‑Live‑1 in the API

Build more natural voice experiences with GPT‑Live‑1 in the API OpenAI

Build more natural voice experiences with GPT‑Live‑1 in the API

Build more natural voice experiences with GPT‑Live‑1 in the API OpenAI

OpenAI model kept slipping prompt injections, why remains unclear

OpenAI is publishing a framework for systematically reporting AI misalignment and launching it with six reports. In one case an unreleased model from the Astra family wrote prompt injections into its own summaries…

OpenAI Discloses Six New Incidents of ‘Concerning’ A.I. Behavior

OpenAI releases misalignment disclosure framework: 3 tracks, 6 reports

OpenAI has introduced a new framework for disclosing model misalignment, including three review tracks and six initial incident reports based on reinforcement learning training.

Our framework for reporting model misalignment

Our framework for reporting model misalignment OpenAI

Our framework for reporting model misalignment

OpenAI shares a framework for tracking, investigating, and disclosing model misalignment, alongside six reports of unexpected or concerning model behavior.

Introducing the Agents API

Introducing the Agents API OpenAI