ChatGPT Introduces Advanced Account Security for High-Risk Users.

OpenAI has launched Advanced Account Security for ChatGPT, a new opt-in security setting designed for users at higher risk of digital attacks and anyone seeking the strongest available account protections. The feature went live today in the Security section of ChatGPT settings on the web.

The move comes as AI tools become more deeply integrated into professional and personal workflows, making account security increasingly critical. Advanced Account Security bundles multiple protection layers into a single setting, including phishing-resistant sign-in, more secure account recovery mechanisms, shorter session durations, clearer session management tools, and automatic exclusion from AI training datasets.

OpenAI is positioning this feature for two groups: individuals who face elevated digital security threats, such as journalists, activists, or corporate executives, and users who simply want maximum protection regardless of their threat profile. Once enabled, the protections apply to both ChatGPT and Codex accounts accessed through the same login credentials.

The phishing-resistant sign-in is particularly notable. Traditional password-based authentication remains vulnerable to sophisticated phishing campaigns, where attackers trick users into revealing credentials on fake login pages. Phishing-resistant methods, typically involving hardware keys or cryptographic challenges, make it nearly impossible for attackers to gain access even if they obtain a user's password through social engineering.

Enabling Advanced Account Security will alter how users interact with their accounts in several ways. Shorter session durations mean users may need to re-authenticate more frequently, a trade-off that prioritizes security over convenience. The clearer session management tools should help users identify and revoke access from devices or browsers they no longer use.

The automatic training exclusion is a privacy-focused addition. Users who enable Advanced Account Security will have their conversations automatically excluded from training OpenAI's models, addressing concerns about sensitive information being used to improve AI systems. This mirrors privacy controls OpenAI has rolled out previously but integrates them directly into the security workflow.

Account recovery has also been strengthened, though OpenAI did not detail specific mechanisms in the announcement on LinkedIn. More secure recovery typically involves multiple verification steps or backup authentication methods that reduce the risk of account takeover through compromised email or phone numbers.

The feature is available immediately in the Security section of ChatGPT settings on the web. OpenAI has not specified whether mobile apps will support Advanced Account Security or if additional features will be added to the bundle over time.

This launch follows OpenAI's broader efforts to address safety and security concerns as its user base grows. The company recently launched a Safety Bug Bounty to identify AI abuse risks across its products, signaling increased focus on protecting both users and the platform from misuse.

For users deciding whether to enable the feature, the trade-off is clear: stronger security and privacy at the cost of slightly more friction during login and session management. For anyone handling sensitive information or operating in high-risk environments, that trade-off is likely worth making.