Technology · Open for opinions
Will new critical security vulnerabilities in large AI models be publicly disclosed by researchers by March 2027?
On 10/18/2026, OBIA reported that Carnegie Mellon University and the Center for AI Safety have already found ways to bypass AI model… 81 opinions, 88% Yes
- Yes — 88%
- No — 12%
Deadline: March 31, 2027 at 12:00 AM UTC · 81 opinions
Resolution rules
On 10/18/2026, OBIA reported that Carnegie Mellon University and the Center for AI Safety have already found ways to bypass AI model security, indicating a continuing trend of offensive research in the field.
Source: obia.nic.br
Frequently asked questions about this event
How is this event resolved?
Resolução afirmativa se pelo menos uma pesquisa acadêmica peer-reviewed ou relatório técnico de instituição reconhecida (ex.: universidades, centros de pesquisa como Carnegie Mellon ou Center for AI Safety) demonstrar publicamente métodos de contorno (jailbreak) de modelos de IA de grandes empresas (OpenAI, Meta, Google ou Anthropic) até 31/03/2027, verificável em repositórios como arXiv ou publicações científicas indexadas.
How are points scored?
Points depend on the share of votes your side had when you gave your opinion. Being right earns 100 minus that share. Being wrong costs that share.
Right now Yes has 88% of the votes: an opinion on it that turns out right earns about 12 points, a wrong one costs about 88.