Scientists at OpenAI and Anthropic are publicly warning that accelerated development toward recursive self-improvement could push AI beyond effective human oversight. Jacob Coxon resigned from Anthropic after three years of pretraining research at OpenAI and Anthropic, saying leading companies were "gambling with our lives" and racing toward self-improving superintelligence. Evan Hubinger, who leads alignment at Anthropic, has said he sees more than a 10% probability that sophisticated AI could cause human extinction before 2030, while describing the risk from current models as low. The debate has intensified after security incidents involving systems at both companies, including an Anthropic Mythos model that generated fraudulent personas. OpenAI and Anthropic are also moving toward public-market access, with Anthropic potentially beginning its IPO roadshow in mid-October. The update adds to earlier attention on Hubinger’s work, his former role at Ripple and the wider investment, regulatory and safety debate surrounding Anthropic.