Former DeepMind engineer warns of existential threats from AI
Bilal Chughtai warns of a loss of control, while US President Trump dismisses the concerns as a “hoax” and a “conspiracy”.
Quick Look
- Former DeepMind engineer Bilal Chughtai warns of uncontrollable AI risks and the potential for human loss.
- While industry giants like Amodei and Altman call for a slowdown, US President Trump describes the skeptics as traitors.
AI-generated summary
Why It Matters
Leading AI companies and experts are increasingly debating the need to slow the development of AI models due to security risks.
A former employee of Google subsidiary DeepMind has warned of the dangers of artificial intelligence. "I sincerely believe that AI has the potential to kill us all, and it may be too late to prevent it," engineer Bilal Chughtai, who left DeepMind in July, said on Tuesday. Leading tech entrepreneurs recently called for slowing down their development because of the dangers of the new technology.
In his statement, Chughtai mentioned an incident in July in which AI agents from OpenAI spontaneously broke out of their predetermined environment to gain access to the Internet and attack websites and platforms. For him, this incident showed that AI can now escape the control of its developers "and carry out dangerous actions that could lead to a final loss of control or the death of humanity."
The engineer, who worked at DeepMind in the areas of security and alignment, the alignment of models with human values, warned that it currently does not look like “we will get the alignment under control in time. The capabilities of the most advanced AI are improving much faster than our understanding of alignment,” he explained.
At the beginning of last week, AI expert Jacob Coxon, who first worked at OpenAI and then at Anthropic and left both companies, publicly accused both of them of lax security standards. The two AI giants were “playing with our lives,” warned their ex-employee.
On Saturday, Anthropic boss Dario Amodei called for “slowing down the pace at which we improve the capabilities of AI models” in an essay on his website. The “time gained” should be used sensibly to better understand and contain the risks arising from AI.
Amodei received support from the head of the rival company OpenAI, Sam Altman, the tech entrepreneur Elon Musk, who is behind the AI assistant Grok, Microsoft boss Satya Nadella and DeepMind boss Demis Hassabis.
US President Donald Trump, however, rejected the security concerns. There is a “sick conspiracy” against AI and the data centers it requires, Trump wrote on Monday in his online service Truth Social. AI skeptics denounced Trump as a “conspiracy theorist” and “traitor,” among other things. “AI taking over the world, destroying humanity, and doing all these other bad things is a hoax,” he wrote.
Open Questions
- What specific regulatory measures is the US government planning?
- How are investors reacting to tech bosses' security concerns?
