- Artificial intelligence researcher Jacob Coxon has resigned from Anthropic, warning that AI firms are gambling with human lives in a race to build self-improving superintelligence.
- Coxon, who also worked at OpenAI, cautioned that superhuman systems could soon possess the ability to hack anything, acquire power, and potentially destroy humanity by the end of the decade.
- He claimed that OpenAI staff have not internalised the civilisational risks, while Anthropic is recklessly pursuing superintelligence to beat competitors to the milestone.
- Anthropic's AI safety lead Evan Hubinger agreed with Coxon's concerns, estimating a greater than 10 per cent chance that AI could kill all humans within the next decade.
- Coxon advocated for urgent cross-industry action to prevent an out-of-control AI race, suggesting a temporary ban on improving model capabilities may be required.
IN FULL
