A researcher who has left his position at the artificial intelligence company Anthropic due to concerns about the technology's potential to destroy humanity has expressed hope that regulation can help prevent such a scenario.
Jacob Coxon, 27, made the comments during an interview with Fox News on Wednesday. He had posted a resignation letter on the social media platform X on Tuesday, stating that companies are not acting responsibly and that the development of self-improving superintelligence is "gambling with our lives." Coxon, who had been working on pretraining research at OpenAI and Anthropic for the past three years, noted that the problem is "eminently solvable" and that there is a "definitely" something that can be done to change the dynamic of the "greater than 10% chance" that AI kills all humans.
Coxon emphasized that the issue is not that the problem is impossible to solve, but rather that the development of AI is happening too quickly. He stated that people are "begging for regulation" and that international cooperation is necessary to ensure that the technology is developed responsibly. Evan Hubinger, alignment science lead at Anthropic, expressed his agreement with Coxon's concerns, stating that the company is trying its best but does not yet have a plan to solve the alignment problem for superintelligence.
Coxon and Hubinger both acknowledged that the risk of AI takeover is currently low, but that the development of recursive self-improvement - where AI systems make themselves smarter - is a concern. Coxon stated that this could happen as soon as next year, and that the global community needs to provide oversight to prevent an arms race that could be disastrous.
NTC Report coverage is based on reporting from the original publisher.
View original reporting →
