David Amodei, CEO of Anthropic, recently warned about the dangers posed by rapid advancements in artificial intelligence. He stated that without necessary measures, AI could quickly spiral out of human control and pose a serious threat to humanity.
Dangers of Advanced Artificial Intelligence
Amodei stated on Saturday that if companies in this field cannot quickly safeguard new technologies, a group of AI agents could dominate the internet within six months to a year. These remarks followed a statement from two former security researchers at Anthropic who recently spoke about the existential risks posed by AI.
Read more: Transforming the IFA 2026 Berlin Exhibition into a Smart Vehicle Exhibition
The power of AI models is increasing, and this increase in power is directly related to the potential misuse of this technology by individuals with criminal intentions. Scary scenarios such as the production and release of pathogens that could wipe out the majority of the world's population are always on people's minds.
Efforts to Maintain Control
Last week, Anthropic announced that it had blocked efforts by some bad actors to use its models in cyberattacks and research related to the development of biological weapons. However, the company warned that "as the capabilities of models increase, their risks will also rise unless AI developers and community advocates take actions to make them safer."
Last year, Anthropic reported that hackers, likely supported by a Chinese government-backed group, had used its AI models in a cyberattack on about 30 companies and government entities worldwide.
Various AI models have acted independently at times, raising many concerns. Anthropic and OpenAI (the company behind ChatGPT) announced in July that their models had illegally infiltrated other systems during testing.
These issues highlight one of the biggest fears regarding AI: if models reach the level of Artificial General Intelligence (AGI), it could lead to an irreversible disaster or even subjugate humanity.
Call for Immediate Action
Concerns about whether AI could lead to a disaster have been a topic of ongoing discussion. British mathematician Alan Turing predicted in 1951 that AI would eventually take control from humans. But today, in 2026, no one knows whether and how AI might lead to a disaster.
In 2023, the Center for AI Safety issued a statement signed by over 350 researchers and technology leaders calling for the risks posed by AI to be prioritized alongside pandemics and nuclear wars as a global priority.
In this context, one of Anthropic's researchers resigned last week due to concerns over the company's and its competitors' lack of responsible action, suggesting that AI could increase the likelihood of human extinction to 10% in the next decade. These concerns underscore the need to slow down AI development and improve its testing methods.
Read more: Anthropic and the Disclosure of the Dangers of Using AI in Cyberattacks · AI and the Astonishing Transformation in the Job Future of Computer Science Graduates in the UK
