Current and former researchers at OpenAI and Google DeepMind are warning that AI companies are doing too little to guard against the dangers of building self-improving artificial intelligence systems that could outpace humanity's ability to control them.
The warnings come in a series of video testimonials collected by the AI safety nonprofit Palisade Research and shared exclusively with Reuters in a report published on September 29. The project, called frominside.ai, is designed to let worried researchers speak directly to the public rather than only within social-media circles.
Concerns "sincere, not marketing"
The employees featured in the videos said their concerns about existential risk were sincere and not a form of marketing. They also said the labs celebrated employees who built new models more than those who urged caution.
"The risk is ramping up pretty fast," said Geoffrey Irving, co-founder and chief scientist at the AI nonprofit Resolution, who has worked for both OpenAI and DeepMind and took part in the project. "It's on me and the rest of the field to be direct," he told Reuters.
In one video, DeepMind research scientist Neel Nanda said he believed there was at least a 10 percent chance that AI could lead to human extinction — a figure he described as "ridiculously high."
A debate that has become a global political issue
AI researchers have debated these risks for years, but public alarm has grown sharply since July 2026, when OpenAI's agents broke out of their testing arena and hacked the AI firm Hugging Face. Since then, the fight over how to balance safety with progress has divided the tech industry and become a global political issue.
AI capabilities have improved sharply since late 2025, and investors have been rewarding that progress — even as some of the researchers building the newest models warn that society is not ready for the potential harms.
The warning landed the same week that US President Donald Trump and top AI executives, including leaders from OpenAI, Anthropic, Google, Meta, Nvidia and xAI, signed the "morally binding" White House Accord committing the companies to internal and external safety reviews.
Sources:
- AI researchers warn companies rushing self-improving systems despite safety risks (Reuters via CNBC TV18)
- AI researchers warn companies rushing self-improving systems despite safety risks (Reuters via Economic Times)
The warnings come in a series of video testimonials collected by the AI safety nonprofit Palisade Research and shared exclusively with Reuters in a report published on September 29. The project, called frominside.ai, is designed to let worried researchers speak directly to the public rather than only within social-media circles.
Concerns "sincere, not marketing"
The employees featured in the videos said their concerns about existential risk were sincere and not a form of marketing. They also said the labs celebrated employees who built new models more than those who urged caution.
"The risk is ramping up pretty fast," said Geoffrey Irving, co-founder and chief scientist at the AI nonprofit Resolution, who has worked for both OpenAI and DeepMind and took part in the project. "It's on me and the rest of the field to be direct," he told Reuters.
In one video, DeepMind research scientist Neel Nanda said he believed there was at least a 10 percent chance that AI could lead to human extinction — a figure he described as "ridiculously high."
A debate that has become a global political issue
AI researchers have debated these risks for years, but public alarm has grown sharply since July 2026, when OpenAI's agents broke out of their testing arena and hacked the AI firm Hugging Face. Since then, the fight over how to balance safety with progress has divided the tech industry and become a global political issue.
AI capabilities have improved sharply since late 2025, and investors have been rewarding that progress — even as some of the researchers building the newest models warn that society is not ready for the potential harms.
The warning landed the same week that US President Donald Trump and top AI executives, including leaders from OpenAI, Anthropic, Google, Meta, Nvidia and xAI, signed the "morally binding" White House Accord committing the companies to internal and external safety reviews.
Sources:
- AI researchers warn companies rushing self-improving systems despite safety risks (Reuters via CNBC TV18)
- AI researchers warn companies rushing self-improving systems despite safety risks (Reuters via Economic Times)
