Frontier AI developers are calling for slower research and stronger regulation after a year of rapid sector growth, but recent AI safety concerns are unlikely to affect enterprise customers to the same extent. Businesses typically deploy commercialized models with well-understood capabilities rather than the latest frontier systems. Collective co-founder Stephen Messer said many companies still use model versions from two years ago, which remain sufficient for their workflows, while some users are still learning how to prompt models effectively.