Will a major AI lab (OpenAI, DeepMind, Anthropic) officially pause the training of a next-generation model for safety reasons for at least 6 months before 2028?

AI Safety Pause: 6 Months Delay by 2028?

Calls for an 'AI Pause' have been frequent. This prediction asks if a major lab will *voluntarily* and *officially* pause the training of a frontier model (post-GPT-5 level) for safety reasons for at least 6 months before 2028.

The Arms Race Dynamic

The majority 'No' vote reflects the commercial and geopolitical arms race. Pausing for 6 months allows competitors (and rival nations) to catch up. While labs talk about safety 'preparedness frameworks', they are incentivized to interpret safety warnings leniently to keep shipping. A pause is more likely to come from government mandate or a lack of compute/energy, rather than a purely voluntary safety stand by a single company.