{"id":623161,"date":"2026-06-05T19:01:08","date_gmt":"2026-06-05T19:01:08","guid":{"rendered":"https:\/\/www.newsbeep.com\/uk\/623161\/"},"modified":"2026-06-05T19:01:08","modified_gmt":"2026-06-05T19:01:08","slug":"anthropic-says-the-world-should-have-option-to-pause-on-ai-ai-artificial-intelligence","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/uk\/623161\/","title":{"rendered":"Anthropic says the world should have option to \u2018pause\u2019 on AI | AI (artificial intelligence)"},"content":{"rendered":"

Anthropic has floated the idea of a worldwide \u201ctemporary pause\u201d on AI development \u2013 and said it was going to convene \u201cpolicymakers\u201d to discuss the dangers of advanced AI \u2013 in its latest release<\/a> touting the capabilities of its products.<\/p>\n

In a long post on Thursday, Anthropic detailed the progress of its AI model, Claude, towards \u201crecursive self-improvement\u201d \u2013 that is, being able to make better and more powerful versions of itself. Recursive self-improvement is a bugbear of AI safety researchers, viewed as the key step for AI to become superintelligent and therefore unleash widespread consequences on humanity.<\/p>\n

The idea features heavily in the widely read AI 2027 doomsday scenario<\/a> of last year, which imagines AI agents designing more and more intelligent versions of themselves, one of which eventually kills all of humanity with a bioweapon in order to make room for more datacentres and solar panels.<\/p>\n

Anthropic\u2019s post notes a \u201ctrend\u201d of increasing capability in Claude which, \u201ctaken far enough and given enough compute \u2026 points to an AI system capable of fully autonomously designing and developing its own successor\u201d. This, Anthropic said, may increase the risk of \u201chumans losing control over AI systems\u201d.<\/p>\n

To deal with this, Anthropic proposed to organise conversations where \u201cpolicymakers, researchers, civil society and other AI companies can help answer some of the questions this piece raises\u201d.<\/p>\n

The news comes alongside a separate report, from the Financial Times, that the US AI company has embedded engineers inside<\/a> the National Security Agency despite a legal battle with the Pentagon over the use of its tools. The engineers are reportedly helping the NSA use Anthropic\u2019s model Mythos for offensive cybersecurity operations.<\/p>\n

If calling for a worldwide conversation on AI risk is in contradiction with supporting a US spy agency to \u2013 potentially \u2013 attack Iran and China with cyberweapons, neither development was \u201csurprising\u201d given the AI company\u2019s past actions, said Steven Murdoch, a professor at University College London.<\/p>\n

\u201cAnthropic might give the impression of being warm and fuzzy, but their definition of AI safety is narrow. Supporting US authorities in the development of offensive capabilities has never been something they have spoken against,\u201d he said.<\/p>\n

Murdoch said Anthropic\u2019s post did not offer evidence of any step changes in the progress of AI capabilities.<\/p>\n

\u201cIt is true that there\u2019s some evidence that AI capabilities have increased and continue to increase with no limits becoming immediately clear,\u201d he said, but he added: \u201cI don\u2019t think anything has fundamentally changed today that has caused Anthropic to publish this article.\u201d<\/p>\n

The advances that Anthropic appears to detail in its post do not amount to AI recursively improving itself \u2013 at least, not yet. Instead, the company reports that a substantial amount of the work done on improving its AI systems is now done with AI. Claude is good at \u201crunning experiments\u201d, it says, or at least speeding certain sections of code.<\/p>\n

Like other AI systems, Claude appears to be improving at tackling more challenging tasks. Anthropic described it \u201csteering research\u201d and \u201cproposing its own experiments\u201d, although these accomplishments appear to have taken place within strict confines, and to be confined to coding-related tasks.<\/p>\n

The quality of code written by AI was also improving, said Anthropic: \u201cAs of May 2026, more than 80% of the code we merge into Anthropic\u2019s codebase was authored by Claude.\u201d<\/p>\n

Murdoch said that Anthropic\u2019s call for a \u201ctemporary pause\u201d on AI echoed<\/a> other proposals on AI safety the company had made throughout the years \u2013 as did its plan to engage policymakers. \u201cIt\u2019s a reminder of what they are concerned about, and have been concerned about for many years.\u201d<\/p>\n

\u201cI\u2019m sure the attention is welcome, but again this isn\u2019t a new thing. Anthropic have been trying to get the attention of policymakers since they were founded.\u201d<\/p>\n

Two months ago, Anthropic announced \u2013 but declined to release \u2013 Mythos, an AI model that it implied was too powerful to unleash upon the public, owing to cybersecurity concerns. The announcement led to widespread hubbub and earned the attention of the US treasury secretary<\/a>, Scott Bessent, and MI5.<\/p>\n

Some experts, however, suggested that there could be more hype behind Anthropic\u2019s Mythos announcement than substance, especially given the vagueness with which the company described some of Mythos\u2019s capabilities. Heidy Khlaaf, the chief AI scientist at the AI Now Institute, called<\/a> the announcement of Mythos \u201ca marketing post\u201d.<\/p>\n

Anthropic this week filed for an IPO that could value the company at $1tn. The company was approached for comment.<\/p>\n","protected":false},"excerpt":{"rendered":"Anthropic has floated the idea of a worldwide \u201ctemporary pause\u201d on AI development \u2013 and said it was…\n","protected":false},"author":2,"featured_media":623162,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[554,733,4308,86,56,54,55],"class_list":["post-623161","post","type-post","status-publish","format-standard","has-post-thumbnail","category-artificial-intelligence","tag-ai","tag-artificial-intelligence","tag-artificialintelligence","tag-technology","tag-uk","tag-united-kingdom","tag-unitedkingdom"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts\/623161","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/comments?post=623161"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/posts\/623161\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/media\/623162"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/media?parent=623161"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/categories?post=623161"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/uk\/wp-json\/wp\/v2\/tags?post=623161"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}