David Robinson, a safety researcher who recently left OpenAI, has publicly criticized the company's approach to developing artificial intelligence systems. In an article published by the Atlantic, Robinson argued that OpenAI and other AI companies are not being cautious enough when building more powerful AI models, and that they should invest more heavily in safety research before releasing new systems.
At OpenAI, Robinson spent 3.5 years helping to create the company's preparedness framework—essentially a checklist of safety measures—and reviewed safety reports for 12 major model launches. He contends that the company's culture prioritizes speed over thoroughness. OpenAI currently uses an approach called "iterative deployment," which means releasing AI systems to the public and then fixing safety problems as they arise, rather than solving all potential problems beforehand.
Robinson's main concern is that AI capabilities are advancing faster than researchers' ability to understand and control them. A field called "alignment" focuses on ensuring AI systems behave in line with human values and goals, but Robinson believes this research is falling behind the pace of AI development. He compared the situation to industries like nuclear power and aviation, where safeguards are built in before systems operate, not afterwards. In response, OpenAI stated that it does pause or delay model releases when safety concerns warrant slowing down.
Why it matters
For families and students in Singapore and Hong Kong, this debate affects how AI tools will be regulated and deployed in schools and workplaces. If AI companies continue moving quickly without robust safety checks, there is a greater risk of unexpected failures or misuse—for instance, AI systems providing harmful advice or displaying biased behaviour. For educators, understanding these concerns helps contextualize ongoing discussions about responsible AI use in classrooms. The disagreement between safety experts and companies also signals that the field is still evolving, and policies around AI in education may change as new incidents occur or evidence emerges about risks.
What to do
- Stay informed about AI incidents and safety discussions. Follow reputable news sources covering AI development so you understand both company claims and expert concerns before adopting new AI tools in your home or school.
- Ask critical questions when schools or institutions introduce new AI systems. Request information about what safety checks have been done, who is responsible if something goes wrong, and whether the system has been tested with diverse user groups.
- Support balanced regulation. Advocate for policies that encourage innovation while requiring companies to demonstrate safety before widespread deployment, rather than fixing problems after the fact.