Version history

OpenAI's GPT-5.5 System Card was updated in an update detected on August 20, 2026. The change involved removing references to three previously listed company and product announcements (Expanding Daybreak as the Cyber Defense Window Narrows, Pacing model development in an era of cyber-critical capabilities, and Introducing ChatGPT for Teens) and adding a new announcement link titled 'Offering Zero Data Retention for frontier models' dated August 19, 2026. This is an editorial change to the card's navigation and related-content section, reflecting current publication links rather than a change to the substantive safeguards or deployment terms described in the system card itself.

OpenAI's GPT-5.5 System Card was detected as updated on August 19, 2026. The change modified the featured content links in the document's header section, replacing three dated articles (published August 10, August 1, and July 29, 2026) with two newer articles published August 18, 2026, including new content on 'Pacing model development in an era of cyber-critical capabilities' and 'Introducing ChatGPT for Teens'. The System Card itself, which documents safeguards for GPT-5.5 and GPT-5.5 Pro deployment, remains substantively unchanged; only the curated links to related resources were updated.

OpenAI's GPT-5.5 System Card, last substantively updated on April 24, 2026, was modified on August 11, 2026. The change replaced a reference to a July 28, 2026 publication ('Scientific computing in the age of agentic AI') with a reference to an August 10, 2026 security publication ('Expanding Daybreak as the Cyber Defense Window Narrows'). This is a minor change to the 'Keep reading' section that directs readers to related OpenAI publications and does not modify any substantive safeguard disclosures or capability descriptions in the System Card itself.

OpenAI's GPT-5.5 System Card was updated in an update detected on August 2, 2026. The change replaced one research publication link in the card's header with a different publication. Specifically, the link to 'Safety and alignment in an era of long-horizon models' (dated July 20, 2026) was removed and replaced with 'Ten advances in mathematics and theoretical computer science' (dated August 1, 2026). This is an editorial update to the featured research content displayed on the system card and does not modify the substantive safeguards documentation or deployment guidance for GPT-5.5.

OpenAI updated the GPT-5.5 System Card detected on July 30, 2026. The change replaced three featured research and safety articles in the document's header with two different research publications. The prior version linked to articles on 'Safety and alignment in an era of long-horizon models' (Jul 20), 'Why teens deserve access to safe AI' (Jul 16), and 'GPT-Red: Unlocking Self-Improvement for Robustness' (Jul 15). The updated version now links to 'How enabling two settings tripled our scores on the ARC-AGI-3 benchmark' (Jul 29) and 'Scientific computing in the age of agentic AI' (Jul 28), while retaining the safety alignment article. This is a navigation and featured content update with no material change to the system card's substantive governance or safety disclosures.

OpenAI's GPT-5.5 System Card was updated in an update detected on July 21, 2026. The document removed a reference to an article titled 'GPT-Red: Unlocking Self-Improvement for Robustness' published July 15, 2026, and added a new reference to 'Safety and alignment in an era of long-horizon models' published July 20, 2026. This change updates the list of related safety resources linked from the system card to reflect the most recent OpenAI safety research publications.

OpenAI updated the GPT-5.5 System Card on April 24, 2026, to include safeguards information. In an update detected on July 18, 2026, a single sentence was modified in the card's reference section. The change replaced a link to one safety-related article (Separating signal from noise in coding evaluations, July 8) with a different safety article (Why teens deserve access to safe AI, July 16). This affects which recent safety publications are highlighted in the system card's navigation.

OpenAI's GPT-5.5 System Card was updated in an update detected on July 16, 2026. The change involved reorganizing the document header and updating the related publications section. Specifically, a new publication titled 'GPT-Red: Unlocking Self-Improvement for Robustness' dated July 15, 2026 was added to the safety research section, replacing previous reference items. This reflects OpenAI's ongoing documentation of safety-related research and system updates.

The OpenAI GPT-5.5 System Card had its title formatting modified in an update detected on July 15, 2026. The document previously opened with 'GPT-5.5 System Card | OpenAI April 23, 2026 Safety Publication' and the updated version removes the leading 'GPT-5.5 System Card |' text, beginning instead with 'OpenAI April 23, 2026 Safety Publication'. This is a minor editorial or formatting change with no material operational impact on the content, governance provisions, or disclosures within the System Card itself.

OpenAI's GPT-5.5 System Card was updated on July 11, 2026 to modify the list of related research and product announcements displayed at the bottom of the card. The previous version linked to a July 8 research article on coding evaluations as the most recent item; the updated version replaces that link structure with references to an OpenAI Bio Bug Bounty Safety announcement dated July 9, 2026. This change updates the associated content and dates but does not alter the substantive safeguard information, deployment guidance, or technical safety documentation contained in the System Card itself.

OpenAI updated the GPT-5.5 System Card on July 9, 2026 by modifying the related research links displayed alongside the card. The updated version replaced two previously listed research articles with a newly published article titled 'Separating signal from noise in coding evaluations' dated July 8, 2026. This change appears to be a routine update to keep featured research current rather than a modification to the system card's substantive safety, deployment, or governance information.