AI Could "Kill All Humans"

By Tech & AI Dispatch

Published: September 15, 2026

A high-profile former artificial intelligence safety researcher from Google DeepMind has sounded the alarm on catastrophic existential risks, warning that humanity could be rapidly running out of time to prevent advanced AI systems from causing total annihilation.

Bilal Chughtai, a research engineer who previously worked on Artificial General Intelligence (AGI) safety and alignment at Google DeepMind until July 2026, published a stark public exit statement across X and LinkedIn on Monday.

"I recently resigned from Google DeepMind, where I worked on AGI safety and alignment research," Chughtai stated. "I earnestly believe that AI has the potential to kill us all, and that we might be running out of time to avoid this outcome."

Chughtai, who co-authored technical safety papers during his tenure at Alphabet’s frontier AI lab, underscored that having observed frontier development firsthand heightened his fears over the industry’s default trajectory.

"A Manic Race": What Prompted the Warning?

Chughtai’s intervention comes at a pivotal juncture for generative and frontier AI systems, where commercial pressures and competitive races between hyperscalers have intensified dramatically.

According to Chughtai, avoiding disaster remains technically possible, but only if the industry puts an end to unregulated, high-speed deployment:

• Slowing Down the Clock: "We need to pace AI development to a speed that society can handle, where emerging risks can be addressed before extreme harm is realised," he emphasized.

• Ending the Competitive Spiral: He stressed that multi-lab coordination is vital to curb the "manic race between AI companies" prioritizing rapid model releases over rigorous alignment verification.

Google did not immediately provide comment on the statement outside of standard business hours.

A Wave of Whistleblowers and Surprising Industry Alignment

Chughtai is far from the lone voice raising existential concerns in recent days. His departure echoes another abrupt exit last week:

• Anthropic’s Jacob Coxon: Resigned from Anthropic with an even more explicit timeline, cautioning that unchecked commercial incentives could precipitate human extinction before 2030. Coxon’s warnings received public support from other former DeepMind scientists, including Alexander Turner.

• Amodei's Call to Slow Down: Over the weekend, Anthropic CEO Dario Amodei called for a controlled deceleration in the release pace of frontier systems.

• A Rare Alliance: Amodei’s call received unexpected consensus from across Silicon Valley, drawing rare supportive commentary from OpenAI CEO Sam Altman and xAI/Tesla chief Elon Musk—reigniting intense global debate around AI existential risk ("doomerism") versus accelerationism.

Political Headwinds and Regulation Stalemates

While alignment engineers and lab leaders urge caution, regulatory action faces steep pushback on the global stage.

In the United States, political polarization continues to cloud legislative efforts to mandate pre-deployment safety audits. Former President Donald Trump recently dismissed tech executives' and researchers' calls for mandatory safety standards on social media, categorizing industry appeals for government guardrails as a "hoax" designed to suppress domestic technological dominance.

Why This Matters Now

These developments signal a critical shift in the AI landscape. Frontline technical researchers closest to alignment and red-teaming evaluations are increasingly willing to break corporate norms and speak directly to the public, underscoring a deep collective anxiety inside top labs. Yet without binding legal frameworks or verifiable cross-industry safety compacts, individual companies remain trapped in a competitive prisoner's dilemma, unable to decelerate unilaterally. As a result, the frontier debate has decisively moved beyond whether human-level AGI is achievable to a far more urgent question: can democratic institutions and regulators audit and govern these systems quickly enough to avert systemic catastrophic failure?

As frontier labs continue training the next generation of reasoning and autonomous agent models, the divide between commercial acceleration and existential precaution has never been starker. Whether Chughtai’s warnings foster meaningful inter-lab safety treaties—or get drowned out in the race to market dominance—remains the defining question for the tech sector.

Connect

contact@blowpost.com

© 2025. All rights reserved.

BlowPost is an independent geopolitical and market intelligence publication. We cater to traders, bankers, logistics professionals, and anyone who wants to understand the structural forces beneath global events — energy markets, capital flows, and trade corridors. No ideology. No institutional interest. Just the structure.

Letters to the Editor :