A former Google DeepMind researcher has issued a stark warning about the potential dangers of artificial intelligence, adding to a growing group of AI scientists who believe increasingly powerful systems could pose an existential risk to humanity.
Bilal Chughtai, who previously worked as a research engineer at Google DeepMind on AGI safety and AI alignment, said he believes artificial intelligence has the potential to cause catastrophic harm and that humanity could be running out of time to prevent such an outcome. Chughtai left Google DeepMind in July 2026.
His comments come amid an increasingly intense debate over the speed of AI development, the race between major technology companies and whether existing safety measures are sufficient to control increasingly capable AI systems.
Former Google DeepMind Researcher Raises Alarm Over AI Safety
Chughtai’s warning has attracted attention because of his background in AI safety and alignment research.
In a statement shared publicly, he said he had recently resigned from Google DeepMind and expressed his belief that AI could potentially “kill us all.” He also argued that society may be running out of time to avoid such an outcome.
Importantly, the warning represents Chughtai’s assessment of a potential future risk rather than evidence that current AI systems are capable of causing human extinction.
His comments nevertheless highlight a growing concern among some researchers- as AI systems become more autonomous and capable, humans may eventually struggle to understand, predict or control their behavior.
Why AI Researchers Are Concerned About Human Control
One of the central issues in the AI safety debate is AI alignment.
AI alignment broadly refers to the challenge of ensuring that powerful AI systems behave according to human goals, values and intentions.
Today’s AI systems remain limited in many important ways. However, researchers are increasingly focused on what could happen if future systems become capable of performing complex tasks with significantly less human supervision.
A particular concern is recursive self-improvement.
This refers to a hypothetical situation in which an advanced AI system contributes to creating or improving subsequent AI systems. If such systems became capable of accelerating their own development, researchers worry that technological progress could move faster than human institutions and safety processes can respond.
A number of researchers have recently highlighted this possibility as one reason for taking AI safety more seriously.
Chughtai’s Warning Follows Similar Concerns From Anthropic Researchers
The former Google DeepMind researcher’s comments come shortly after warnings from researchers associated with Anthropic.
Anthropic researcher Jacob Coxon recently resigned and said people developing advanced AI systems genuinely believe the technology could potentially kill humanity by the end of the decade.
Anthropic scientist Evan Hubinger subsequently said he agreed with Coxon’s concern and estimated that there could be a greater than 10% chance of AI causing human extinction within the next decade.
Such estimates are highly uncertain and remain the subject of significant disagreement within the scientific and technology communities.
Nevertheless, the fact that researchers working inside leading AI organizations are publicly discussing catastrophic AI risks has intensified calls for stronger safety measures.
AI Development Race Adds to Safety Concerns
One of the biggest concerns raised by AI safety researchers is the growing competition between major technology companies.
Companies such as OpenAI, Anthropic and Google DeepMind are competing to build increasingly capable AI models and autonomous AI agents.
The commercial incentives are enormous. More capable AI systems could transform software development, scientific research, business operations and other industries.
However, critics argue that intense competition could create pressure to release increasingly powerful systems before researchers have fully understood their risks.
Chughtai specifically called for greater coordination between AI companies, arguing that safely developing the technology would require avoiding what he described as a race between companies.
Dario Amodei Calls for a Slower Pace of AI Development
The latest warnings also coincide with calls from Anthropic CEO Dario Amodei for a slowdown in advanced AI development.
Amodei has argued that the industry should give safety measures more time to catch up with increasingly powerful AI systems.
His position has received unusual support from several prominent technology figures, including OpenAI CEO Sam Altman and Elon Musk, although there remains substantial disagreement over how AI development should be regulated and whether slowing development is practical.
The debate highlights a fundamental problem facing the AI industry- companies want to move quickly enough to remain competitive, while safety researchers want sufficient time to evaluate increasingly powerful systems.
What Is AI Alignment and Why Does It Matter?
AI alignment has become one of the most important areas of artificial intelligence safety research.
The basic goal is to ensure that an AI system’s objectives remain consistent with what humans actually want.
This becomes more complicated as AI systems become more autonomous.
A simple chatbot responding to a question is relatively different from an AI agent capable of writing software, operating digital systems, conducting research and making decisions independently.
As AI agents gain more autonomy, researchers need to understand how they behave when given complex objectives and how they respond when those objectives conflict with human instructions.
The challenge becomes even greater if future systems become significantly more capable than humans in certain domains.
Could AI Really Kill Humanity?
There is currently no scientific consensus that AI will inevitably cause human extinction.
The claim that AI could kill all humans is a risk scenario discussed by some researchers, not an established prediction.
Experts disagree substantially about the probability of such an outcome.
Some AI researchers believe advanced AI could create unprecedented risks if systems become highly autonomous and difficult to control. Others argue that extinction predictions are too speculative and that policymakers should focus more heavily on immediate AI-related problems such as misinformation, cybersecurity, job displacement and misuse.
Even researchers who take existential risks seriously generally emphasize that the outcome is not predetermined.
The objective of AI safety research is precisely to identify dangerous behaviors and develop mechanisms that allow humans to maintain meaningful control over advanced systems.
Why Recursive Self-Improvement Is Receiving Attention
Recursive self-improvement has become a major topic in discussions about future AI risks.
The concept describes a potential cycle in which AI helps researchers build better AI, which then becomes capable of making even more advanced systems.
If this process accelerates dramatically, technological development could potentially occur at a pace that humans find difficult to monitor.
Researchers do not currently know whether truly autonomous recursive self-improvement will occur in the way described by some scenarios.
However, the possibility is one reason AI safety researchers are studying how future systems might behave when given greater autonomy.
Former Google DeepMind researcher Rishub Jain, for example, has also raised concerns about humans gradually removing themselves from the development process as AI becomes increasingly capable of helping build successor systems.
AI Safety vs. AI Innovation
The latest warnings have also created a broader debate about how society should balance AI innovation and AI safety.
Supporters of rapid development argue that artificial intelligence could deliver enormous benefits in medicine, science, education, productivity and economic growth.
Slowing development too aggressively could also create problems if competitors continue advancing their systems.
On the other hand, AI safety researchers argue that certain risks could become much harder to address after highly autonomous systems are deployed at scale.
This creates a difficult policy question- How fast should humanity develop technology that could eventually become more capable than humans in important areas?
There is no simple answer.
Calls for Greater Regulation and International Cooperation
Several AI researchers believe that individual companies cannot solve advanced AI safety problems alone.
Chughtai has called for coordination between AI developers so that companies are not forced into a race where safety considerations are sacrificed for speed.
This has contributed to wider discussions about government oversight, independent AI safety organizations, model evaluations and international cooperation.
Some policymakers are also beginning to examine whether existing regulatory systems are sufficient for increasingly powerful AI technologies.
The challenge is designing rules that reduce genuine risks without preventing beneficial AI research and development.
AI Industry Faces Growing Pressure
The latest warning comes at a critical moment for the AI industry.
Technology companies are spending enormous amounts on computing infrastructure, data centers and advanced chips while investors expect AI to generate substantial economic returns.
At the same time, researchers are increasingly questioning whether the industry is moving too quickly.
The combination of enormous financial investment and growing safety concerns could make AI regulation one of the most important technology policy debates of the coming years.
For companies, the challenge will be demonstrating that they can continue innovating while maintaining credible safety standards.
What Happens Next?
The debate surrounding AI’s long-term risks is unlikely to disappear.
As models become more capable, researchers will continue testing their ability to reason, write software, operate autonomously and complete complex tasks.
Governments will also face pressure to establish rules for the development of advanced AI systems.
Meanwhile, technology companies will have to decide how much emphasis to place on speed, competition and commercial opportunity versus safety research and risk management.
The warnings from former Google DeepMind and current or former Anthropic researchers do not prove that AI will cause human extinction. However, they demonstrate that some people with direct experience developing advanced AI systems believe the potential risks deserve serious attention.
Conclusion- AI Safety Becomes More Urgent as Technology Advances
The warning from former Google DeepMind researcher Bilal Chughtai adds to a growing conversation about the potential existential risks of artificial intelligence.
His concerns, alongside warnings from researchers at Anthropic and other AI organizations, highlight fears that the race to build increasingly capable systems could move faster than humanity’s ability to guarantee their safety.
Whether the most extreme AI scenarios ever materialize remains uncertain. What is clear is that AI safety, alignment and regulation are becoming increasingly important as the technology advances.
For the AI industry, the challenge is no longer simply building more powerful systems. It is also ensuring that those systems remain useful, predictable and controllable as their capabilities grow.
Kode Digital Blog is a technology-driven platform sharing insights on software development, AI, digital marketing, and business growth.
© 2026 Kode Digital. All rights reserved.
