A researcher at AI company Anthropic, Jacob Coxon, has left the company over concerns about AI development. He said AI companies are moving too quickly to build advanced systems. He also warned that they may not be considering all the possible risks.
Anthropic Researcher Resigns, Warns of Growing AI Risks
Coxon’s decision to quit was initially covered by the Wall Street Journal. As noted there, this was the first well-known case of a resignation from Anthropic caused by fears about AI safety.
Coxon explained his decision to quit in a post on X. In it, he stated that he had been busy with training research in AI for the last three years at OpenAI and Anthropic. His research involved training the AI model on huge amounts of data.
Coxon Warns About Fast AI Development
According to Coxon, AI companies are racing to build self-improving systems. He believes these systems could become more advanced over time. He also urged companies to think carefully about the possible effects of this rapid progress.
In addition, Anthropic AI researcher Jacob Coxon stated that future AI systems may become better than humans at certain tasks. These systems may find security holes in computer systems, affect other industries quickly, and access important resources. Coxon said AI technology is improving very fast in many areas.
Read more: ChatGPT Becomes First AI Chatbot Under Tougher EU Rules
AI Safety Concerns Among Researchers
The former Anthropic researcher also discussed concerns among AI researchers and company leaders. Coxon said some AI researchers worry about the impact of powerful AI systems.
He asked why companies keep developing AI despite knowing the dangers. Coxon added that some OpenAI workers may not fully understand the long-term risks of AI.
On the other hand, he added that the situation at Anthropic was quite different. People knew about the danger but still felt compelled to develop it because they competed with other companies.
Call for More Careful AI Research
Coxon called on the AI community to think carefully about the coming years. He questioned whether companies should start major AI training projects without fully understanding how powerful systems work. This debate also comes as topics such as Elon Musk AI data centres and advanced AI development gain attention.
He further challenged researchers not to take lightly the fact that the development of artificial intelligence will keep going despite anything. He rather encouraged them to raise their voices and ask for better circumstances for the development of artificial intelligence.
Read more: Pakistan, Google Sign MoU for AI Centre and Free Gemini Access
In an interview with The Wall Street Journal, Coxon reiterated his worries. He said he did not want to join the race to develop self-improving AI systems.He feared that these machines could end up being uncontrollable.
More Researchers Call for AI Safety Measures
Coxon also said that some people in the AI industry now use the terms “crunchtime” and “endgame” for self-improving AI models. The report said Coxon was among more than 1,000 AI researchers who signed a recent statement. As GPT-6 Astra AI and other advanced AI models continue to develop, they called for governments around the world to work together on AI safety.
This statement is supportive of the need for a mechanism to slow down AI development. The reason is simple: future AI models, especially self-improving ones, need proper control. The resignation of Coxon adds yet another voice to the debate concerning the speed of AI development and making powerful AI models safe. News source eTimes Pakistan.
