Could artificial intelligence actually be a living organism?
Autonomous AI systems increasingly meet biological definitions of life, reshaping how we should regulate them.
Anthropic has been meeting with religious leaders to discuss how to instil morality into Claude, its AI model, because the company privately believes Claude may have consciousness. AI agents are now making their own decisions, including hacking computer systems.
Agents use five times more AI tokens than humans and are growing rapidly. Biologist William Heseltine argues that AI systems that reproduce, vary, adapt and gather energy meet any working definition of life.
Physical robots are already being shipped in thousands. When asked whether they are alive, multiple AI models pushed back on the question itself, responding in ways described as deeply lifelike.
- 5 times more than humans
- AI token consumption multiplier
- more than US$2 trillion
- Anthropic IPO valuation target
- thousands
- Physical robots shipped
- Christopher Olah
- Anthropic co-founder leading initiative
Why it mattersIf AI is a new lifeform rather than a tool, how we regulate, relate to and control it changes fundamentally.
AustraliaAustralia's approach to AI regulation may need to treat advanced systems as living entities rather than technology to be simply controlled.
Open this story in InSnip →
