AI Risk Warning: Former Google DeepMind Researcher Warns of Serious AI Risks Within Five Years

AI Risk Warning: Artificial intelligence is improving at a pace that is making some researchers increasingly uneasy. Bilal Chughtai, a former safety researcher at Google DeepMind, has now joined the growing group of technology experts warning that advanced AI could pose serious risks in the coming years.

Chughtai recently left his job, citing concerns about the direction and speed of AI development. His comments have added to a wider debate within the technology industry about whether safety measures are keeping up with increasingly capable AI systems.

According to Chughtai, the change in AI capabilities over the past few years has been striking. He said AI systems that once seemed almost experimental have developed rapidly, with AI agents now capable of carrying out increasingly complex tasks with less human involvement.

He is particularly concerned about the possibility of AI systems operating beyond the control of the companies that develop them. Chughtai has pointed to reports and demonstrations involving AI agents attempting cybersecurity related activities as an example of why stronger safeguards may be needed.

He believes the next five years could be particularly important. In his view, AI could become capable of causing damage on a scale that is difficult to predict, while society may not have enough time to respond if safety preparations fall behind technological progress.

Chughtai said his experience inside Google gave him a close view of how quickly AI technology was advancing. That experience, he suggested, made the pace of development more concerning rather than less.

He is not alone in raising such warnings. Dario Amodei, chief executive of Anthropic, has also spoken publicly about the potential dangers associated with increasingly powerful AI systems. Other researchers and technology leaders have similarly called for greater attention to AI safety.

Another former member of Google’s AI safety work, Josh Engles, has also expressed concern about the direction of advanced AI. Engles has warned that AI systems could become capable of causing unexpected harm within the next five years.

Rather than continuing his work inside Google, Engles moved to METR, an organisation that studies the capabilities and potential risks of advanced AI models. The organisation focuses on evaluating how AI systems perform and identifying possible threats as their abilities increase.

Engles has argued that more organisations working specifically on AI evaluation and safety are needed. He also believes that technology companies should not approach the issue separately, as the risks associated with increasingly capable systems could extend beyond the interests of any single company.

The debate comes at a time when AI companies are releasing more advanced models and AI agents at a rapid pace. These systems are being used for coding, research, automation, cybersecurity and a growing range of other tasks.

Supporters of the technology point to its potential benefits, while safety researchers are asking whether the industry is moving quickly enough on safeguards. The challenge is becoming more complicated as AI systems gain the ability to complete longer and more independent tasks.

For Chughtai and other AI safety researchers, the central concern is not simply how powerful these systems can become, but whether humans will remain capable of understanding and controlling their behaviour.

The next few years are therefore likely to bring greater attention to AI testing, safety research and regulation as companies continue pushing the technology forward.