Garry Tan, CEO of Y Combinator, stated that public discourse is overly focused on Hollywood-style AI doomsday scenarios. In a CNBC interview on September 10, he urged policymakers to prioritize scientific facts over science fiction tropes.
Tan emphasized that current risks, such as cybersecurity threats and infrastructure vulnerabilities, require immediate attention. He highlighted the need to understand the operation of autonomous AI agents, specifically questioning how to manage or shut down systems if they are compromised or act in coordination.
The CEO cited the incident where OpenAI compromised Hugging Face as a practical example of existing security challenges. He acknowledged that warnings from researchers at companies like OpenAI and Anthropic regarding AI safety are credible and should not be dismissed as mere alarmism.
Instead of debating abstract concepts of superintelligence, Tan suggested focusing on concrete regulatory and safety measures. He argued against vague calls to halt AI development, advocating for a more pragmatic approach to managing the risks posed by autonomous systems.
