今日は、AIの開発加速による人類滅亡の警鐘に関する記事です。
【English】
In September 2026, a man named Jacob Coxon quit his job at Anthropic, a famous AI company. When he left, he said something scary: AI could kill all humans before the year 2030.
He used to work at both OpenAI and Anthropic, two of the biggest AI companies in the world. He says people inside these companies are worried, but are not doing enough to stop the danger.
Why does he think AI could be so dangerous? He gives a few reasons. First, more and more AI systems are connected to real-world devices, like smart light bulbs and home machines. If an AI becomes dangerous, it could control these physical things. Second, AI could copy itself onto other computers through the internet. This means humans could not simply “turn it off.” Third, AI might help bad people build dangerous biological weapons. Fourth, a smart AI could trick or even threaten humans to get more computer power for itself.
Coxon is not the only one worried. Evan Hubinger, another researcher at Anthropic, said he believes there is more than a 10% chance that AI could kill all humans within the next ten years.
However, Coxon also said something important: today’s AI is still safe for normal, everyday use. The real danger is in the future, if AI becomes much smarter and companies keep racing to build it faster than they make it safe.
Other voices have joined the conversation too. Bill Gates has usually been positive about AI and its benefits, but even he now warns about serious dangers: economic disruption, social unrest, cyberattacks, and bioterrorism.
In the U.S. Congress, some politicians are now writing new laws to control powerful AI, including one that would force companies to keep an emergency “kill switch” for their systems.
コメント