AI leaders are talking about decelerating development for safety and security, a sign the AI existential risk needs to be critically addressed
Following Jacob Coxon’s resignation from Anthropic and his remarks to the BBC that people working in the industry are “genuinely frightened” by the pace of development, Anthropic CEO Dario Amodei proposed industrywide deceleration and stronger oversight, with OpenAI’s Sam Altman and Elon Musk of xAI both publicly backing that direction.
Amodei outlined three global alignments: slowing AI development, imposing global regulation, and creating independent monitoring of advanced systems.
This concurrence among leaders of competing AI companies signals that concerns about the pace and oversight of AI development have moved firmly into the industry mainstream.
“The people sounding the alarm on untamed AI have more visibility into these systems than almost anyone else, so their warnings deserve to be taken seriously,” commented Laura Ellis, Senior Vice President of Artificial Intelligence, Arctic Wolf.
But while whether AI becomes an existential threat years from now is an important debate, she believes it is not something individuals or corporations can build safeguards against right now.
What then do these warnings from AI leaders mean for organisations trying to manage AI risk today?
“What they can safeguard are the gaps that already let AI systems exceed their permissions, reach sensitive data, or operate outside the controls meant to contain them,” said Ellis.
These gaps “are the same problem the alarm-raisers are describing, at a scale we can see and manage today: systems doing more than we intended, with too little oversight.”
The Singapore situation
Arctic Wolf’s 2026 AI & Cybersecurity Trends Report
echoed the same concern, with data pointing to AI and agentic AI risks representing the top three cybersecurity concerns for Singapore organisations.
Ellis added: “This year, Singapore’s IMDA also released a dedicated governance framework for agentic AI, precisely because the risks of systems acting outside their intended boundaries have become real.”
She advised that AI agents should be managed like any other privileged identity: “…With clear guardrails, human oversight for high-impact decisions, and the ability to revoke access immediately when something goes wrong.”
“No organisation can eliminate every risk. The ones that succeed with AI are those that continuously reduce exposure, detect issues early, and recover quickly enough that a single failure does not become a lasting business problem.”
