OpenAI releases
OpenAI news, ChatGPT, models and product updates. New features, breaking changes, security advisories and deprecations - each summarised in plain English and updated continuously.
Tracking 162 OpenAI releases · Updated
- OpenAI News aisecurity ·
OpenAI: PRC-linked influence operations target AI debates in US
OpenAI's new report reveals influence operations linked to the People's Republic of China are leveraging AI to target U.S. discussions on technology, data centers, and tariffs. These operations are also disseminating false claims about OpenAI's products, including ChatGPT. The findings highlight the increasing use of AI in geopolitical information warfare.
announcement - OpenAI News aisecurity ·
OpenAI Election Information and Safeguards
OpenAI is enhancing access to election information, bolstering cyber defenses, and improving AI transparency in anticipation of global elections. These measures aim to support users and defenders during the upcoming election cycles. The focus is on information accessibility and AI integrity.
announcement - OpenAI News aisecurity ·
ChatGPT enhances safety with improved context recognition
OpenAI has updated ChatGPT to better recognize context in sensitive conversations, improving its ability to detect risk over time and respond more safely. This enhancement aims to provide a more secure user experience for all participants. The update is now available for users engaging in sensitive discussions.
feature announcement - OpenAI News aisecurityengineer ·
OpenAI Responds to TanStack npm Supply Chain Attack
OpenAI has detailed its response to the TanStack "Mini Shai-Hulud" supply chain attack, outlining measures taken to secure systems and signing certificates. macOS users are required to update OpenAI applications by June 12, 2026, to maintain security. The announcement explains the incident, its impact, and OpenAI's enhanced defenses against future software supply chain threats.
security announcement - OpenAI News aisecurityengineer ·
OpenAI's security measures for running Codex
OpenAI has implemented multiple security layers, including sandboxing, approval workflows, network policies, and agent-native telemetry, to ensure the secure and compliant operation of its Codex coding agent. These measures are designed to facilitate the safe adoption of AI coding assistants within the organization. The approach focuses on robust protection for code generation and execution.
announcement - OpenAI News aisecurity ·
OpenAI Expands Trusted Access for Cyber with New GPT Models
OpenAI has expanded its Trusted Access program for cybersecurity with the introduction of GPT-5.5 and GPT-5.5-Cyber. These models aim to help verified defenders accelerate vulnerability research and enhance protection for critical infrastructure. This expansion is particularly relevant for organizations focused on cybersecurity defense and threat intelligence.
announcement feature - OpenAI News security ·
OpenAI Enhances Account Security with New Features
OpenAI has introduced Advanced Account Security, a new feature set designed to protect user accounts and sensitive data. This includes phishing-resistant login methods and improved account recovery options to prevent unauthorized access. The enhancements are aimed at all OpenAI users to safeguard against account takeover.
feature security - OpenAI News aisecurityarchitectgovernment ·
OpenAI's Cybersecurity Plan for the Intelligence Age
OpenAI has released a five-part action plan detailing strategies for enhancing cybersecurity within the context of the Intelligence Age. The plan emphasizes democratizing AI-driven cyber defense capabilities and focuses on safeguarding critical infrastructure. This initiative aims to leverage AI to protect vital systems against evolving threats.
announcement - OpenAI News aisecurityengineer ·
OpenAI details ChatGPT community safety measures
OpenAI has outlined its strategies for ensuring community safety within ChatGPT, focusing on model safeguards, misuse detection, and policy enforcement. These efforts involve collaboration with external safety experts to maintain a secure user environment. The company emphasizes its ongoing commitment to responsible AI development and deployment.
announcement - OpenAI News aisecuritycompliancegovernment ·
OpenAI Services Achieve FedRAMP Moderate Authorization
OpenAI has achieved FedRAMP Moderate authorization for both ChatGPT Enterprise and the OpenAI API. This authorization is crucial for U.S. federal agencies looking to securely adopt advanced AI technologies. It signifies that OpenAI's services meet stringent federal security requirements for cloud computing.
announcement - OpenAI News aisecurityengineer ·
GPT-5.5 Bio Bug Bounty Launched
OpenAI has launched the GPT-5.5 Bio Bug Bounty, a red-teaming initiative focused on discovering universal jailbreaks related to biological safety risks. Participants can earn rewards up to $25,000 for successful findings. This challenge aims to proactively identify and mitigate potential security vulnerabilities in AI models.
announcement - OpenAI News aisecurityengineergovernment ·
OpenAI Cyber Defense Ecosystem Adds Security Firms with GPT-5.4-Cyber
OpenAI is launching the Trusted Access for Cyber program, incorporating leading security firms to leverage GPT-5.4-Cyber. This initiative aims to bolster global cyber defense capabilities, supported by $10 million in API grants. The program seeks to accelerate the development of advanced security solutions by trusted partners.
announcement feature - OpenAI News aisecurityengineergovernment ·
OpenAI Expands Cyber Program with GPT-5.4-Cyber
OpenAI has expanded its Trusted Access for Cyber program by introducing GPT-5.4-Cyber for vetted cybersecurity defenders. This expansion aims to strengthen safeguards as AI capabilities in cybersecurity continue to advance. The new model is available to a select group of cybersecurity professionals.
feature announcement - OpenAI News aisecurityengineer ·
OpenAI addresses Axios developer tool compromise
OpenAI has responded to a supply chain attack originating from Axios by rotating macOS code signing certificates and updating applications. The company confirmed that no user data was compromised during the incident. This response aims to secure their development environment and reassure users about data privacy. Engineers working with OpenAI tools on macOS should be aware of these security measures.
security announcement