![](\"https:\/\/fortune.com\/img-assets\/wp-content\/uploads\/2026\/06\/GettyImages-2261864825-e1781368061354.jpg?w=2048\")
<\/p>\n<\/p>\n
Research that caused the Commerce Department to impose strict limits on the use of Anthropic\u2019s new AI models wasn\u2019t geared toward offensive purposes, according to a cybersecurity CEO who saw the findings.<\/p>\n
Late Friday, the department used national security export controls to bar the company from distributing\u00a0Fable 5 and Mythos 5. <\/p>\n
The directive applies to people outside the U.S. and foreign nationals in the U.S., including Anthropic\u2019s own non-citizen employees. Due to the directive\u2019s scope, Anthropic said it had no choice but to disable the models for all users.<\/p>\n
The company said it was told that research on a \u201cjailbreak\u201d of Anthropic\u2019s AI that sought to probe bypassing of safeguards sparked the export controls. <\/p>\n
\u201cWe disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people,\u201d Anthropic wrote in a blog post. \u201cIf this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.\u201d<\/p>\n
While the company reaffirmed the government\u2019s ability to block unsafe AI, it argued that should be part of a statutory process that\u2019s transparent fair, and based on technical facts. \u201cThis action does not adhere to those principles.\u201d<\/p>\n
Katie Moussouris, CEO of cybersecurity firm Luta Security, told the Wall Street Journal<\/em> that Anthropic showed her a copy of the findings, which were produced by\u00a0Amazon researchers using prompts to obtain information about security vulnerabilities.<\/p>\n \u201cI\u2019ve seen the paper. It\u2019s not a jailbreak. It was Defense Oriented Prompting (DOP), capabilities defenders need,\u201d she explained in a post on X on Saturday. <\/p>\n Moussouris added, \u201cIf Nat defense is the goal, this just scored an own goal against us.\u201d<\/p>\n The Commerce Department didn\u2019t immediately respond to requests for comment.<\/p>\n \u201cAs a leading cloud provider that serves a large number of private and public sector customers, it\u2019s not uncommon for governments to seek our counsel on potential security risks. When they occur, we don\u2019t share the details of these discussions,\u201d an Amazon spokesperson told Fortune<\/em>.<\/p>\n Meanwhile, the administration\u2019s directive barring foreign nationals in the U.S. from using Anthropic\u2019s new models raised alarms.<\/p>\n Ben Murphy, a scholar at the Institute for Progress think tank, called it \u201canother step on the balkanization of technology.\u201d<\/p>\n \u201cIt might previously have been unthinkable to require proof-of-citizenship to access services; it\u2019s increasingly common across new technologies and, to be honest, this attempt is not surprising in that light,\u201d he added in a post on X. <\/p>\n Murphy also highlighted the unpredictability of the administration\u2019s actions and its consequence for AI developers, warning that labs could keep more models in-house or not make them available.<\/p>\n In addition, labs might be less inclined toward engaging with the government about potential vulnerabilities, he said, with Anthropic\u2019s stance on being transparent seeming to backfire.<\/p>\n Anthropic was already feuding with the administration, which has deemed it a supply-chain risk for Pentagon contractors. Still, the company provided early access to the Mythos model as it warned on its potential cybersecurity implications.<\/p>\n \u201cI don\u2019t know that the government wouldn\u2019t have reached that conclusion themselves, but as a business matter, those pronouncements have not produced a healthy working relationship with the government,\u201d Murphy wrote.<\/p>\n<\/div>\n #Research #leading #restrictions #Anthropic #models #wasnt #jailbreak #cybersecurity #CEO<\/p>\n","protected":false},"excerpt":{"rendered":" Research that caused the Commerce Department to impose strict limits on the use of Anthropic\u2019s new AI models wasn\u2019t geared toward offensive purposes, according to a cybersecurity CEO who saw… <\/p>\n","protected":false},"author":1,"featured_media":7151,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[536,369,337,6774,9503,2965,2453,195,9502,6405],"class_list":["post-7150","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-finance-news","tag-anthropic","tag-ceo","tag-cyber","tag-cybersecurity","tag-jailbreak","tag-leading","tag-models","tag-research","tag-restrictions","tag-wasnt"],"_links":{"self":[{"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=\/wp\/v2\/posts\/7150","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=7150"}],"version-history":[{"count":0,"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=\/wp\/v2\/posts\/7150\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=\/wp\/v2\/media\/7151"}],"wp:attachment":[{"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=7150"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=7150"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.fintechpulse8.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=7150"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}