OpenAI

Monitors coding agents for misalignment and security risks.

Introduces context compaction for long-running agent tasks.

Enables command line interactions for complex agent tasks.

Equips Responses API with a computer environment for agents.

Advanced reasoning model with cybersecurity safety mitigations.

Newest GPT-5 series model with faster, richer responses.

Standardized risk labels for certain ChatGPT capabilities.

Advanced security setting for high-risk ChatGPT users.

Advanced coding model with enhanced reasoning and execution capabilities.

Program offering $10 million in API credits for cybersecurity.

Custom AI agent for data exploration and reasoning at OpenAI.

Prevents data leaks through URLs in AI agent interactions.

Core logic orchestrating user, model, and tool interactions.

Testing ads in ChatGPT for free and Go tiers.

Low-cost subscription tier offering expanded AI capabilities.

AI-enhanced code review tool for identifying systemic risks.

Advanced agentic coding model for complex software engineering tasks.

Expanding ChatGPT Enterprise to all 120,000 BBVA employees globally.

Advisory group for cybersecurity and frontier capability domains.

Tiered access to enhanced AI capabilities for cyberdefense.

Agentic security researcher for finding and fixing vulnerabilities.

Course for teachers to integrate AI in classrooms.

Hands-on training for practical AI skills across industries.

Evaluation measures model performance on real-world tasks.

State of the art video and audio generation model.