OpenAI and Yubico are partnering to bring custom YubiKeys to OpenAI users for Advanced Account Security, providing phishing-resistant, hardware-backed authentication for ChatGPT accounts.

AAS replaces password logins with stronger methods like passkeys and hardware keys, significantly reducing credential theft risk.

AAS is optional and targets high‑risk groups—journalists, researchers, officials, and others handling sensitive work—while remaining available to all users for added protection.

The move occurs within the evolving AI cloud security landscape and is discussed alongside broader AI deployment and security coverage.

TechCrunch reported the feature, with a public announcement on May Day 2026 and foundations laid in late April 2026.

Authored by Lucas Ropé of TechCrunch, the piece frames AI security within rising threat scenarios.

While beneficial for high‑risk users, the security enhancements may be most impactful for conversations involving sensitive material, underscoring broader security implications for AI providers.

If a security key is lost, recovery becomes challenging, so OpenAI emphasizes planning and policy for secure recovery and access management.

Adoption requires careful recovery and access policy planning, since losing a key can complicate account continuity.

The rollout aligns with industry-wide moves toward phishing-resistant authentication rather than responding to a single breach, aiming to bolster security amid growing AI use.

In India, a major market with about 100 million weekly active ChatGPT users, the program targets high‑risk users like journalists, researchers, and officials amid documented security threats.