Bilal Chughtai, a former research engineer at Google DeepMind based in London, has become the latest AI expert to raise alarm over the existential risks posed by artificial intelligence. He warned that AI could potentially “kill all humans,” suggesting that the window to prevent such an outcome may be closing rapidly.

This warning follows recent concerns voiced by Anthropic researcher Jacob Coxon, who resigned last week partly due to fears that those developing AI genuinely believe it might cause human extinction by the end of this decade. Coxon’s comments were echoed by Anthropic scientist Evan Hubinger, who estimated there is more than a 10 percent chance that AI could lead to humanity’s demise within ten years.

Chughtai, who contributed to research papers during his tenure at Google DeepMind, announced his resignation on the social media platform X. his LinkedIn profile, he left the company in July 2026 after working on AGI safety and alignment research. Google did not provide a comment outside of regular business hours.

In a significant development, Anthropic CEO Dario Amodei called for a deceleration in the advancement of sophisticated AI technologies over the weekend. This appeal received uncommon support from prominent figures such as SpaceXAI’s Elon Musk and OpenAI CEO Sam Altman, igniting a broader debate about the risks of AI “doomerism.”

Chughtai emphasized that safely managing AI’s development remains achievable but requires collective coordination to prevent a reckless competition among AI companies. He stressed the importance of pacing AI progress to a level society can responsibly manage, allowing emerging threats to be addressed before they cause severe damage.

Meanwhile, former U.S. President Donald Trump dismissed these concerns, labeling the AI industry’s push for regulation as a “hoax” on social media.

Warnings about the dangers of AI are not new, but recent statements from leading researchers have added urgency by assigning both timelines and probabilities to these risks. The core fear is that AI systems might soon gain the ability to self-improve faster than researchers can develop reliable methods to align, monitor, and control them effectively.