Article
【Andrew Ng: AI dangers are being overhyped; development should not be paused】
Stanford University professor and Landing AI founder Andrew Ng wrote on X that recent voices promoting AI dangers have made significant headway, apparently driven by a carefully planned public-relations campaign, and that he worries this represents a setback for the field. Ng said AI technology has not taken an unexpected dangerous turn, that the risk of human extinction has not increased relative to several months ago, and that the related theories remain science-fiction scenarios. He noted that the largest change is in cybersecurity capabilities, which should be taken seriously but will not cause the end of the world. A recent incident in which an OpenAI team deployed a swarm of agents to breach Hugging Face was excessively sensationalized by the media; for example, 1200 agents working in parallel were portrayed as a magical capability, whereas computers already run large numbers of processes simultaneously. Vulnerabilities in OpenAI's sandbox and monitoring were the key factors; fixing the vulnerabilities and strengthening monitoring would suffice, with no need to pause AI. Ng also said that large language models and agents should not be anthropomorphized: responsibility lies with their builders and users, not with the tools themselves. Pausing AI progress would cause more harm, as competitors will not slow down; engineering should identify and fix problems through empirical evidence. Beneficial applications still far outweigh risks, and development should continue.