AI BriefingOpenAI Blog news
The latest AI and developer news from OpenAI Blog, with short summaries.
Visit website →OpenAI Introduces Safety Bug Bounty Program
OpenAI has unveiled a 'Safety Bug Bounty' program to identify AI misuse and safety risks.
Helping developers build safer AI experiences for teens
OpenAI has released a prompt-based safety policy package to help developers build safe AI services for teens.
OpenAI Foundation Update
The OpenAI Foundation will invest at least $1 billion to address disease treatment and economic impact.
ChatGPT Enhances Product Discovery Features
OpenAI is leveraging the Agentic Commerce Protocol (ACP) to strengthen visual and conversational product discovery and shopping features within ChatGPT.
Safe Creation with Sora
OpenAI announced multi-faceted safeguards for safe video generation on Sora 2 and the Sora app.
How OpenAI Monitors Misalignment in Its Internal Coding Agent
OpenAI has built a monitoring system based on GPT-5.4 Thinking to detect misalignment risks in its internal coding agent.
OpenAI Announces Acquisition of Astral
OpenAI is acquiring Python developer tooling company Astral to expand the Codex ecosystem.
Introducing GPT-5.4 mini and nano
OpenAI has released GPT-5.4 mini and nano, high-performance small models optimized for coding and subagent tasks.
OpenAI Japan Announces 'Japan Teen Safety Blueprint' for Youth Safety
OpenAI Japan has announced the 'Japan Teen Safety Blueprint' to support teenagers' safe use of generative AI.
Providing Workers with Compensation Insights
American workers are using about 3 million ChatGPT messages every day to close the pay information gap.
Why Codex Security Doesn't Provide SAST Reports
Codex Security goes beyond the limitations of SAST, which only tracks data flow, by adopting a method that verifies the mismatch between the intent of the code and its actual behavior.
How to Design AI Agents to Counter Prompt Injection
As prompt injection evolves into social engineering techniques, designs are needed that go beyond simple filtering to limit the system's exposure to influence.
From Models to Agents: Building Computer Environments into the Responses API
OpenAI combines a shell tool with container environments in the Responses API to enable models to act as agents that carry out complex workflows.
Wayfair Improves Catalog Accuracy and Support Speed by Adopting OpenAI
Wayfair integrated OpenAI models into its core workflows to improve catalog data quality and automate its support process.
Rakuten Doubles Incident Recovery Speed by Adopting Codex
Rakuten adopted OpenAI's **Codex** to cut mean time to recovery (MTTR) by 50% and boost project build speed by up to 4x.
Improving the Instruction Hierarchy of Frontier LLMs
OpenAI released IH-Challenge, a reinforcement learning dataset that helps models correctly judge priority among conflicting instructions.
A new way to learn math and science in ChatGPT
ChatGPT has launched an interactive visualization feature that lets users learn math and science concepts by manipulating them in real time.
OpenAI Acquires Promptfoo
OpenAI is acquiring the security platform Promptfoo to strengthen the security and evaluation capabilities of AI agents.
Codex Security: Research Preview Released
OpenAI has unveiled Codex Security, a security agent that understands system context to identify vulnerabilities and even suggest patches.
How Descript Built Large-Scale Multilingual Video Dubbing Technology
Descript leveraged OpenAI's reasoning models to build large-scale multilingual video dubbing technology that simultaneously optimizes meaning and video length.