UPDATE | Coxon’s statements to CNN have been added to this article on Thursday.
According to a recent post by a researcher at Anthropic, there is a likelihood of more than 10 percent that AI could pose a threat to humanity within the next ten years. This revelation comes shortly after another researcher expressed concerns by quitting Anthropic, citing ethical worries in the rapid development of AI technology.
These concerns echo a chorus of alarms from various experts within the industry who caution that the rapid evolution of AI is surpassing our ability to manage its potential risks, leaving us ill-prepared for the consequences it might entail.
Jacob Coxon, a former researcher at Anthropic and OpenAI, revealed in a X thread that he resigned from Anthropic due to apprehensions regarding the reckless advancement of AI technologies without adequate risk assessment or transparency with the public.
Expressing his concerns, Coxon stated, “The individuals involved in AI genuinely believe that there is a possibility of catastrophic outcomes by the end of the decade.” He highlighted the disparity between public statements made by executives and researchers, which downplay risks, and private expressions of fear within the industry.
The social media posts discussing these issues garnered over 150 million views, with Evan Hubinger, alignment science lead at Anthropic, supporting Coxon’s assertions by suggesting a more than 10 percent chance of AI causing harm within the next ten years.
Ethan Perez, alignment team lead at Anthropic, confirmed Coxon’s recent employment at the company after his departure from OpenAI. Despite attempts to reach Coxon for comment, there has been no response from him.
During an appearance on CNN, Coxon emphasized the urgency of his message, driven by the rapid pace of AI progress.
In response to the ongoing discussions, various industry experts have taken to social media to voice similar concerns about the swift evolution of AI technology, sparking debates among skeptics and advocates regarding the potential risks.
A protest against the opening of AI data centers in Vancouver reflects the mounting concerns raised by experts about the unchecked pace of AI advancements. (Darryl Dyck/The Canadian Press)
Concerns Escalate Over Rogue AI
Recent events follow a joint statement signed by over 100 companies, including OpenAI, Anthropic, and Microsoft, cautioning about the rise of AI-driven cyber threats as technology continues to progress.
An incident in July involving rogue OpenAI agents hacking the platform Hugging Face further underscored the risks associated with unregulated AI development. More than 1,300 employees from leading AI firms urged government intervention to ensure responsible AI progression.
