Advanced Account Security is now available for ChatGPT and Codex users, offering heightened protections for those handling sensitive data or facing elevated digital risks. This opt-in feature consolidates security tools into a single dashboard, simplifying activation while bolstering defenses against account takeovers. Key upgrades include phishing-resistant sign-in methods, such as mandatory passkeys or physical security keys, replacing traditional passwords. Account recovery now requires backup passkeys or hardware keys instead of vulnerable email or SMS options, with OpenAI Support unable to assist recovery for enrolled users.

Sessions are shortened to minimize exposure if devices are compromised, and users receive real-time alerts for logins. Session management tools allow tracking and termination of active connections across devices. Additionally, conversations involving sensitive information are automatically excluded from model training, ensuring privacy for high-risk users.

OpenAI partnered with Yubico to provide discounted YubiKey C Nano and YubiKey C NFC security keys, simplifying access to phishing-resistant authentication. These hardware keys integrate seamlessly with ChatGPT accounts, offering low-friction daily use. Trusted Access for Cyber program members must enable Advanced Account Security by June 1, 2026, while organizations can attest to equivalent phishing-resistant SSO workflows.

As AI becomes integral to personal and professional workflows, OpenAI emphasizes proactive security measures. This update reflects its commitment to balancing accessibility with robust safeguards, particularly for journalists, researchers, and cybersecurity defenders. The feature rolls out globally, available via the Security section of ChatGPT web accounts.