AI workers split on existential risk as sceptics dismiss extinction warnings

September 20, 2026 at 04:01 AM2 min

This digest was compiled by AI from multiple sources — links to the originals are below.

Employees at major AI firms including OpenAI, Meta and DeepMind are pushing back on recent warnings that artificial intelligence could endanger humanity. Multiple current and former workers told the BBC they found the claims vague and lacking evidence. The scepticism follows viral warnings from former Anthropic employee Jacob Coxon and others urging a slowdown in AI development.

Key Facts

Multiple employees at OpenAI, Meta and DeepMind told the BBC they were sceptical of claims that unchecked AI development could kill people en masse.
Former Anthropic employee Jacob Coxon said a group of AI agents based on models that do not currently exist could decide to create and aim a biological weapon.
Rishub Jain, founder of AI safety firm Sampura Research and former DeepMind employee, said the tone among AI workers regarding fresh fears was 'definitely a little jokey'.
Meta data scientist Colin Fraser wrote that there was no real evidence AI models would inevitably pursue a goal leading to human death.

Scepticism Among AI Workers

Employees who have worked for OpenAI, Meta and DeepMind expressed doubt about warnings that AI could endanger humanity. A former OpenAI employee who knew Jacob Coxon said their first reaction to his claims was amusement, citing a lack of detail. The person described the existential risk claims as 'always vague', with specific assertions tending toward major jumps in reasoning or hypothetical circumstances. Rishub Jain, who spent seven years at DeepMind before founding Sampura Research, said the current tone among AI workers was 'definitely a little jokey'.

The Coxon Warning

Jacob Coxon, a former Anthropic employee, claimed a group of AI agents based on models that do not currently exist could decide to create and aim a biological weapon. Coxon did not detail how exactly such a scenario would take place. His warning went viral and was echoed by employees of Anthropic, OpenAI, DeepMind and Elon Musk, who runs the AI startup xAI.

Technical Rebuttals

Meta data scientist Colin Fraser wrote on social media that there was no real evidence AI models would inevitably pursue a goal leading to human death. Fraser summarised his technical argument with the phrase: 'LLMs won't wipe out humanity because they just don't have that dog in them.' Rishub Jain noted that people in AI companies have discussed existential risk for many years, so the recent warnings did not come as a sudden shock.