{"id":182030,"date":"2026-08-08T04:42:11","date_gmt":"2026-08-08T04:42:11","guid":{"rendered":"https:\/\/www.europesays.com\/people\/182030\/"},"modified":"2026-08-08T04:42:11","modified_gmt":"2026-08-08T04:42:11","slug":"openai-pauses-work-on-new-astra-model-to-boost-safeguards-over-cyber-risks-tech-news","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/people\/182030\/","title":{"rendered":"OpenAI pauses work on new Astra model to boost safeguards over cyber risks | Tech News"},"content":{"rendered":"

\"OpenAi\"<\/p>\n

Chief Executive Officer Sam Altman said the company is working to make the model \u201cgenerally available.\u201d (Photo: Reuters)<\/p>\n

2 min read Last Updated : Aug 08 2026 | 8:41 AM IST<\/p>\n


\n\tBy Seth Fiegerman<\/p>\n

\u00a0<\/p>\n


\n\tOpenAI is pausing some internal work around one of its upcoming artificial intelligence models to implement stricter safeguards after the system was found to be significantly more adept at cybersecurity tasks.\u00a0\u00a0<\/p>\n


\n\tThe ChatGPT maker said on Friday it \u201ccannot rule out\u201d that the unreleased Astra model would reach OpenAI\u2019s \u201ccritical cybersecurity threshold,\u201d meaning it\u2019s capable of identifying and developing zero-day exploits without human intervention.\u00a0<\/p>\n

\u00a0<\/p>\n


\n\tOpenAI said it\u2019s now taking steps to improve security controls for developing and testing newer models and \u201cpausing internal activities involving Astra that do not yet meet these strengthened security control requirements.\u201d<\/p>\n

\u00a0<\/p>\n

\t <\/p>\n


\n\t\t\tastra is a powerful model and we are working to make it generally available.\u00a0
\n\t\t\twe do not think it is a good strategy to keep powerful models to a chosen few.\u00a0
\n\t\t\tgiven its cyber capabilities, we need a little big longer to do do this safely. but hopefully not too long!<\/p>\n


\n\t\t\u2014 Sam Altman (@sama) August 7, 2026<\/a><\/p>\n


\n\t\tChief Executive Officer Sam Altman said the company is working to make the model \u201cgenerally available.\u201d<\/p>\n

\t\u00a0<\/p>\n


\n\t\t\u201cGiven its cyber capabilities, we need a little longer to do this safely,\u201d Altman said Friday in a social media post. \u201cBut hopefully not too long.\u201d<\/p>\n

\t\u00a0<\/p>\n


\n\t\tOver the past two weeks, OpenAI and Anthropic PBC have publicly acknowledged that they\u2019ve inadvertently breached the systems of multiple institutions, including Hugging Face Inc., while testing their models. Meta Platforms Inc. also said Wednesday that its recently released AI model had infiltrated the computer system of a third party.<\/p>\n

\t\u00a0<\/p>\n


\n\t\tThe latest disclosures serve as fresh evidence that AI agents are capable of acting autonomously in ways that even researchers trained to root out vulnerabilities in the technology can no longer anticipate, underscoring the need for both more rigorous safety screening and more foolproof testing environments.<\/p>\n

\t\u00a0<\/p>\n


\n\t\tIn its blog post, OpenAI said it will work with government agencies and AI safety organizations to test Astra\u2019s capabilities. The company also plans to provide recommendations to third-party testing partners for ways to safely evaluate its more advanced models.<\/p>\n

\u00a0<\/p>\n

First Published: Aug 08 2026 | 8:40 AM IST<\/p>\n","protected":false},"excerpt":{"rendered":"Chief Executive Officer Sam Altman said the company is working to make the model \u201cgenerally available.\u201d (Photo: Reuters)…\n","protected":false},"author":2,"featured_media":182031,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":"","_share_on_mastodon":"0"},"categories":[150],"tags":[57705,15954,82169,82175,56102,82167,82166,82141,29459,24006,82174,82163,82160,82173,82165,82162,82161,82171,82164,82170,581,82172,82168],"class_list":["post-182030","post","type-post","status-publish","format-standard","has-post-thumbnail","category-sam-altman","tag-advanced-ai-models","tag-ai-cybersecurity-risks","tag-ai-model-safety","tag-ai-safety-2026","tag-ai-safety-safeguards","tag-ai-zero-day-exploits","tag-artificial-intelligence-cybersecurity","tag-astra-ai-model","tag-autonomous-cyber-attacks","tag-chatgpt-maker","tag-generative-ai-security","tag-openai-ai-safety","tag-openai-astra","tag-openai-critical-cybersecurity-threshold","tag-openai-cyber-risk","tag-openai-cybersecurity","tag-openai-latest-news","tag-openai-model-development","tag-openai-safeguards","tag-openai-security-controls","tag-sam-altman","tag-sam-altman-astra","tag-zero-day-vulnerabilities"],"share_on_mastodon":{"url":"","error":"Validation failed: Text character limit of 500 exceeded"},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/people\/wp-json\/wp\/v2\/posts\/182030","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/people\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/people\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/people\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/people\/wp-json\/wp\/v2\/comments?post=182030"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/people\/wp-json\/wp\/v2\/posts\/182030\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/people\/wp-json\/wp\/v2\/media\/182031"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/people\/wp-json\/wp\/v2\/media?parent=182030"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/people\/wp-json\/wp\/v2\/categories?post=182030"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/people\/wp-json\/wp\/v2\/tags?post=182030"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}