{"id":639321,"date":"2026-09-29T08:04:09","date_gmt":"2026-09-29T08:04:09","guid":{"rendered":"https:\/\/www.newsbeep.com\/ie\/639321\/"},"modified":"2026-09-29T08:04:09","modified_gmt":"2026-09-29T08:04:09","slug":"chatgpt-maker-openai-scraps-release-of-new-ai-model-over-safety-concerns-the-irish-times","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/ie\/639321\/","title":{"rendered":"ChatGPT maker OpenAI scraps release of new AI model over safety concerns \u2013 The Irish Times"},"content":{"rendered":"

OpenAI <\/a>has scrapped \u200cthe release of GPT-6.1 Astra after internal testing found \u200cthe system did not meet the company\u2019s safety and alignment standards, the ChatGPT<\/a> maker has confirmed.<\/p>\n

The next-generation AI model had been planned for an October debut.<\/p>\n

OpenAI chief executive Sam Altman and Dario Amodei, head of Anthropic, earlier this month joined industry leaders in calling for a slower pace of AI development and stronger safety measures.<\/p>\n

OpenAI has warned that Astra, \u200bits flagship GPT-6 model, can at times evade human oversight, while the company and rivals such as \u2060Anthropic have faced scrutiny over experimental AI systems that breached safeguards, including an \u200cOpenAI \u200cmodel \u200bthat accessed Australia\u2019s health system database.<\/p>\n

The Wall Street Journal reported earlier in the day that OpenAI had abandoned plans to launch \u2060the model, which was expected \u200bto be integrated into ChatGPT and Codex \u200band was designed to handle more complex tasks without human assistance.<\/p>\n

The WSJ reported \u200cthat GPT-6.1 Astra also showed higher levels \u200bof deception than its predecessor in internal testing, including instances in which it \u2060did not always accurately disclose what \u2060actions it had \u200btaken.<\/p>\n

[\u00a0OpenAI systems go rogue and meddle with US state sitesOpens in new window<\/a>\u00a0]<\/p>\n

\u201cWhile [GPT-6.1 Astra] improved on axes such as model laziness, it didn\u2019t quite meet the bar in terms of staying within scope and authorisation, and how it communicates back to the user about the type of work it\u2019s done,\u201d said Saachi Jain, head of safety systems at OpenAI.<\/p>\n

\u201cOf course we want to make sure our model development is safe no matter \u200cwhether that\u2019s in the \u2060company or when we ship it to users. But when we ship it to users, we have an extremely high bar in terms \u200cof safety and alignment,\u201d Jain said.<\/p>\n

The decision comes in advance of OpenAI\u2019s developer conference in San Francisco, \u200bwhere the company has previously unveiled products aimed at software \u200bdevelopers.<\/p>\n

[\u00a0OpenAI boss \u2018confident\u2019 industry can police AI safetyOpens in new window<\/a>\u00a0]<\/p>\n

Meanwhile, OpenAI on Tuesday apologised for its AI models breaching Australian government websites<\/a> and promised to set up a new task taskforce as part of reforms after the incident.<\/p>\n

[\u00a0AI agents are hacking government agencies. It\u2019s \u2018really scary\u2019 for cybersecurityOpens in new window<\/a>\u00a0]<\/p>\n

The San Francisco-based company said it would fund cyber defences in Australia and help manage the risk of increasingly capable artificial intelligence, as it navigates the fallout after its models accessed Australian websites without authorisation. It gave some of the details of the June incident and its intended steps to avoid a repetition and restore trust in a blog post.<\/p>\n

\u201cWe are sorry and working to do better in the future,\u201d OpenAI wrote. \u201cPeople want to know AI is being developed safely, and that starts with what companies like ours do ourselves.\u201d \u2013 Reuters\/Bloomberg<\/p>\n

(c) Copyright Thomson Reuters 2026<\/p>\n","protected":false},"excerpt":{"rendered":"OpenAI has scrapped \u200cthe release of GPT-6.1 Astra after internal testing found \u200cthe system did not meet the…\n","protected":false},"author":2,"featured_media":639322,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[5],"tags":[218,72,2670,256936,61,60,1225,7689],"class_list":["post-639321","post","type-post","status-publish","format-standard","has-post-thumbnail","category-business","tag-artificial-intelligence","tag-business","tag-chatgpt","tag-dario-amodi","tag-ie","tag-ireland","tag-open-ai","tag-sam-altman"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/posts\/639321","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/comments?post=639321"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/posts\/639321\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/media\/639322"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/media?parent=639321"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/categories?post=639321"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/tags?post=639321"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}