AI might actually ‘kill all humans’

[Sevam]

BBC News dropped a chilling report on AI safety. Top Anthropic researcher Evan Hubinger warns that while current models are safe, rapidly developing self-improving AI poses a >10% chance of human extinction within a decade.

Adding fuel to the fire, Jacob Coxon - who just quit Anthropic and used to work at OpenAI—explicitly states that neither company is acting responsibly. The video breaks down their warnings, the industry reaction, and how we can stop this.
Are tech giants moving too fast?

[RobertMiller]

I welcome AI progress, but existential risk estimates deserve serious scrutiny. We need measurable alignment, adversarial testing, human-in-the-loop controls and transparent incident reporting. Neither blind optimism nor doomcasting helps.