AI’s rapid expansion fuels debate over safety, human control and social costs
publishers 158articles 587first reported 23 Sep, 14:00 UTCdeveloping since 30 Sep · 53 editions
Artificial intelligence’s expanding capabilities and infrastructure are prompting renewed debate over who controls the technology, how its risks are assessed and what role remains for human judgment.
OpenAI CEO Sam Altman said at The New York Times DealBook Summit that handing decision-making authority to AI would be a “terrible idea” and that people must remain in control. OpenAI President Greg Brockman said more capable models may require new monitoring methods because they can increasingly perform tasks without writing out an observable chain of thought.
Separately, OpenAI announced a pilot program embedding staff from four outside safety and research organizations to evaluate potential risks. The company said assessors would receive access to sensitive tools and models, but researchers interviewed by NPR argued that the program is narrower than Altman’s earlier promise of “employee-like access,” citing restrictions on assessors’ contact with employees.
Demand for AI computing continues to grow despite recent warnings from leading labs. The Verge reported that Nvidia’s quarterly data-center revenue rose 110% year over year to $107.6 billion, while AMD’s data-center sales nearly doubled to $8.5 billion.
AI’s broader effects are also drawing scrutiny. Cybersecurity experts told CNA that AI-assisted vulnerability discovery is increasing pressure to deploy emergency software patches quickly, potentially contributing to outages when updates receive limited testing. At a Seoul arts forum, artists and academics discussed how AI affects learning and creativity, emphasizing the continuing value of human understanding and experimentation. A climate briefing by analyst Ketan Joshi argued that data-center growth is also driving substantial fossil-fuel demand and emissions.
HOW THIS STORY WAS MADE
Written from 8 articles, 7 with the publisher's own text; 8 independent newsrooms once syndicated copies count as one; this version written 1170 h after the record first saw the story; the editor kept it as written.
- Publishers
- 158
- Source articles
- 587
- Given to the writer
- 8, 7 with the article's own text
- Independent newsrooms
- 8
- This version written
- 1170 h after the record first saw the story
- Second model (editor)
- kept as written
- Publication gate
- passed
- Human review
- none
WHAT THE EDITOR DID
The second model checked the writer's draft against the same sources and published it as written.
Written by a language model from the sources above, then checked by a second model that may only cut, attribute or correct. How it works →