Extract from ABC News
Pete Hegseth will update a directive on the autonomy of weapons systems "to ensure the deliberate adoption of AI systems that respect the chain of command". (Reuters: Evan Vucci)
In short:
The White House said it would accelerate the development and use of AI for national security applications, including "intelligence and war fighting domains".
Major AI lab Anthropic is warning of an increased risk of "humans losing control over AI systems".
The Trump administration has stressed the technology should not be used to carry out unlawful surveillance.
The White House says it will accelerate the development and use of artificial intelligence for national security applications.
The news came as a major AI company warned of an increased risk of "humans losing control over AI systems".
The Trump administration announced the plans while stressing the technology should not be used to carry out unlawful surveillance.
Earlier, it asked leading AI developers to voluntarily submit their most capable models for government cybersecurity tests before releasing them to the public.
"Under my Administration, the United States can and will responsibly accelerate the use of AI across intelligence and war fighting domains in line with American values," US President Donald Trump said in a national security memorandum.
Mr Trump said US Defense Secretary Pete Hegseth had 90 days to update an existing directive on the autonomy of weapons systems "to ensure the deliberate adoption of AI systems that respect the chain of command".
The US president added AI technologies should not be developed or used by the national security enterprise "to censor free speech … or conduct unauthorised or unlawful surveillance activities".
AI increasingly developing other AI systems, Anthropic warns
The Trump administration's plans for AI comes as Anthropic warns companies are increasingly delegating AI development to other AI systems.
In a blog post, the US company said if it was taken far enough, the trend could lead to "an AI system capable of fully autonomously designing and developing its own successor".
This post poses the latest warning from Anthropic, whose co-founder Christopher Olah recently praised Pope Leo for his encyclical on the rise of AI.
As of May 2026, more than 80 per cent of the code merged into Anthropic's coding system was authored by its own chatbot, Claude, it said.
It also said that by 2027 Claude would be capable of tasks that took a person weeks, according to research.
"These trends have huge implications. AI that can build itself would be a major development in the history of technology — one that could bring enormous good for the world in science, healthcare, and beyond," it wrote.
"But full recursive self-improvement also might increase the risks of humans losing control over AI systems.
"If systems are capable of fully building their own successors, the ways we secure them, monitor them, and shape their behavior all grow much more important."
The blog post came in the wake of a clash between Anthropic and the Pentagon.
The Pentagon slapped a formal supply-chain risk designation on Anthropic in March after it refused to back down on bans against Claude being used to power autonomous weapons and mass US surveillance.
The Pentagon said it should be able to use the technology as needed, as long as it complied with US law.
Further in its post, Anthropic said while Claude's code was improving, many of the staff believed "the Claude-written code was still worse in quality than human-written code at Anthropic in late 2025".
With the potential for this to change shortly, it proposed a global "option to slow or temporarily pause frontier AI development to enable societal structures and alignment research to keep up with the advance of the technology".
It said it would work to build systems with other labs to allow this slowdown to occur.
"A meaningful slowdown or pause would require multiple well-resourced labs at or near the frontier, in multiple countries, agreeing to stop under the same conditions, the authors wrote.
"It would also require that each can verify that the others have actually stopped."
A "meaningful slowdown" would require multiple labs to agree to the idea. (Reuters: Dado Ruvic)
It noted the difficulty of this approach, given AI can be "far easier to conceal than missile silos".
"None of this is necessarily impossible in principle — the world has built verification regimes for other complex technologies (e.g., the Intermediate-Range Nuclear Forces Treaty)— but those regimes took decades to build both the infrastructure and the trust" the blog post concluded.
"We don't have that long.
"A unilateral pause by one lab, by contrast, is achievable immediately, but accomplishes much less: it would change who the front-runner is, but it would not create the wider deliberative process that is currently missing."
Reuters/ABC
No comments:
Post a Comment