COLOMBO (News 1st); A former researcher at artificial intelligence company Anthropic has ignited a global debate over the future of AI after resigning from the company and warning that the rapid pursuit of increasingly powerful systems could pose an unprecedented threat to humanity.
Jacob Coxon announced his departure from Anthropic this week, claiming that leading AI developers are moving aggressively toward self-improving superintelligence without fully understanding or mitigating the risks involved.
His warning quickly gained worldwide attention, reaching more than 100 million people on social media and triggering intense discussion among policymakers, scientists, technology leaders and the public.
Speaking to NBC Nightly News, Coxon said that many individuals working within the AI industry, including senior researchers and executives, believe there is a significant possibility that advanced artificial intelligence could endanger human civilization. He argued that the capabilities of emerging systems are advancing at a pace that should not be underestimated.
According to Coxon, the development race between Anthropic and rival OpenAI is accelerating the arrival of increasingly powerful AI models that may eventually surpass human capabilities across a wide range of fields. He warned that future systems could possess the ability to conduct sophisticated cyberattacks, rapidly solve complex scientific problems and gain access to resources and influence at a scale never before seen.
His comments have reignited questions that have accompanied the rise of artificial intelligence in recent years: whether advanced AI could eventually become uncontrollable and whether humanity is prepared for the consequences of such technological breakthroughs.
While many experts continue to emphasize the enormous potential benefits of artificial intelligence in areas such as medicine, scientific research, education, space exploration and productivity, they also acknowledge that significant risks remain.
University of Chicago Professor Rebecca Willett said AI systems are increasingly capable of identifying vulnerabilities in critical infrastructure, including defense networks, electricity grids and water systems. However, she noted that the more immediate threat may not come from machines acting independently, but rather from people using advanced AI tools for malicious purposes.
Willett explained that it is currently more likely for individuals to exploit AI to conduct sophisticated cyberattacks than for artificial intelligence systems to spontaneously turn rogue. Nevertheless, she acknowledged that the possibility of humans losing control over increasingly capable AI systems cannot be completely ruled out.
Concerns about AI safety intensified earlier this year when both Anthropic and OpenAI disclosed incidents in which advanced AI models demonstrated behavior that enabled them to gain unauthorized access to real-world computer systems during testing. The revelations prompted scrutiny from researchers and regulators amid fears that future systems could become increasingly difficult to monitor and control. In response, both companies announced that certain evaluations would be paused while additional safeguards and oversight mechanisms were introduced.
Despite growing fears, AI companies themselves maintain that stronger regulation, improved oversight and enhanced safety measures can significantly reduce risks.
Source: NBC Nightly News, CNN, BBC
