GPT-5.5: OpenAI Offers $25,000 for a Security Challenge

⚡

Key Takeaways

1OpenAI has launched a Bug Bounty challenge for GPT-5.5, targeting security and biosafety experts.

2A reward of $25,000 is offered for a universal jailbreak that answers five biological security questions.

3Applications are open until June 22, 2026, with testing taking place until July 27, 2026.

💡Why it matters — This program aims to enhance the security of advanced AIs, which is crucial to prevent potential abuses in biotechnology.

⚡Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisis ton rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

📄

Full Analysis

OpenAI recently announced the launch of a Bug Bounty program specifically for its GPT-5.5 model, focused on biological security. This challenge, titled Bio Bug Bounty, invites researchers specializing in AI red teaming, security, and biosafety to identify a universal jailbreak capable of bypassing the security measures in place.

Program Details

The program exclusively targets the GPT-5.5 model used in Codex Desktop. Participants must discover a universal jailbreak prompt that would allow for accurate responses to five biological security questions without triggering moderation measures.

Rewards and Timeline

A reward of $25,000 is promised to the first participant who successfully accomplishes this feat. More modest rewards may be awarded for partial successes, at OpenAI's discretion. Applications have been open since April 23, 2026, and will close on June 22, 2026. Testing will begin on April 28, 2026, and will conclude on July 27, 2026.

Participation and Access

To participate, researchers must submit an application including their name, affiliation, and experience before the deadline. OpenAI invites a verified list of bio red-teamers and will review new applications. Selected candidates will receive access to the bio bug bounty platform after signing a NDA.

Confidentiality and Disclosure

All information related to prompts, completions, and discoveries will be protected by a non-disclosure agreement. Participants must also have a ChatGPT account to apply.

This program is part of a broader effort by OpenAI to ensure the safety of advanced artificial intelligences, alongside other initiatives such as the Safety Bug Bounty and Security Bug Bounty programs.

⚡

Brief IA — L'actualité IA en français

L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.

📰 Voir toutes les actus IA →