from a substack: "Our forecasters think there’s a 7% probability (5% to 8%) that OpenAI will have to “de-deploy” GPT-5.5, making it unavailable to the general public, before 2027." It's related to GPT 5.5 being as good as Mythos at cyber per UK AISI and Mythos being held back.
---
This market resolves YES if, before January 1, 2027, OpenAI removes, disables, substantially restricts, or withdraws public access to GPT-5.5 primarily because of safety, security, misuse, cyber-capability, regulatory, national-security, or “too powerful to release” concerns.

This includes cases where OpenAI says or strongly indicates that GPT-5.5 is being withheld, rolled back, gated, suspended, or removed because its capabilities create unacceptable risk, including but not limited to cyber misuse, biosecurity, autonomous-agent misuse, model scheming, persuasion, fraud, or other dangerous capability concerns.

This market resolves NO if GPT-5.5 is removed, deprecated, renamed, hidden, rate-limited, or made unavailable primarily because of normal product lifecycle reasons, including replacement by a newer model, redundancy, cost, infrastructure constraints, product simplification, branding changes, API cleanup, or routine sunset of older models.

If GPT-5.5 is replaced by a newer or stronger model and OpenAI does not attribute GPT-5.5’s removal to safety, security, cyber-risk, misuse, regulatory, or other threat-related concerns, this resolves NO.

If no model branded or clearly recognized by OpenAI as GPT-5.5 is released to the general public before January 1, 2027, this resolves NO.

Primary sources for resolution are official OpenAI blog posts, system cards, preparedness/safety reports, release notes, API documentation, press releases, or official OpenAI social media statements. Credible reporting may be used as supporting evidence only if official statements are ambiguous.

Routine retirement because GPT-5.5 becomes obsolete or is replaced by GPT-6 would not count. The intended YES case is a risk-motivated rollback/withdrawal/restriction, especially around cyber or dangerous capability concerns.

Background

This market is based on speculative forecasting regarding the potential for OpenAI to withhold or withdraw future advanced AI models due to safety, security, or regulatory concerns. The prompt references discussions surrounding the UK Artificial Intelligence Safety Institute (AISI) evaluations and comparisons to other models (e.g., "Mythos"). Traders may wish to monitor OpenAI’s official release communications and statements regarding the safety protocols and public availability of their future model iterations.