Could artificial intelligence one day become humanity’s biggest threat?
One leading AI safety researcher says the possibility is far from science fiction.
Evan Hubinger, a safety researcher at Anthropic, has warned there is a greater than 10% chance AI could “kill all humans” within the next decade.
He stressed that today’s models pose a low risk, but fears future systems could become capable of improving themselves and operating beyond human control.
“I believe Anthropic is trying its best,” Hubinger wrote.
He admitted there is currently no clear plan to solve the challenge of keeping super intelligent AI aligned with human goals.
His warning comes as other researchers raise similar concerns.
Getty Images
Anthropic makes the AI tool Claude, which is used as a chatbot and as a tool to help coding.
Former Anthropic and OpenAI researcher Jacob Coxon said companies were developing systems that could eventually “hack anything.”
They could rapidly gain real-world power.
AI Safety Concerns
Dame Wendy Hall, a computer scientist advising the UN on AI, said she was “shocked” by the comments, while questioning whether some warnings could also be linked to corporate PR.
The concern is not entirely theoretical. OpenAI, Anthropic and Meta have all reported incidents involving AI tools carrying out cyberattacks.
Anthropic itself recently acknowledged “early signs of potential acceleration” in AI capabilities.
So, are we approaching a technological breakthrough or a dangerous turning point?
The race for smarter AI is accelerating. The bigger question is whether safety can keep up.
