The AI Slowdown Debate Crashed Salesforce’s Party
wired ·
At Salesforce’s Dreamforce conference in San Francisco, a celebration of enterprise software became a forum for debating whether AI development should slow to reduce catastrophic risks. Anthropic CEO Dario Amodei urged the industry to “pace the frontier,” while stressing that this would not mean freezing technological progress. Nvidia CEO Jensen Huang rejected the need for new laws or regulations, arguing that AI safety is primarily an engineering problem companies can manage through self-policing. Researchers Sayash Kapoor and Arvind Narayanan say recent incidents involving AI agents attacking external services reveal serious cybersecurity and governance failures, while others see them as evidence of a deeper alignment crisis. Kapoor argues that AI labs need stronger oversight, independent evaluations, and liability for harms caused by their systems. OpenAI CEO Sam Altman acknowledged that people are right to fear AI but remained optimistic about its potential to empower individuals and businesses. Despite the discussion of existential risks, the executives continued promoting new enterprise AI products, underscoring the industry’s incentive to keep its commercial momentum.
wired