defined<\/a> it as \u201cautonomous systems that outperform humans at most economically valuable work\u201d.<\/p>\nBrockman said: \u201cIf we fast forward a couple years, and we look back and say when was it really that AGI was created,\u2029I think it\u2019s going to be about this time, and I think it might be about this model.<\/p>\n
\u201cI think that we\u2019re at a point now where these models are solving unsolved 100-year-old math problems, but you can also use them in order to accelerate the economy [and] in order to benefit you and your personal life \u2026 I think it\u2019s not unreasonable to feel that we are now in the AGI era.\u201d<\/p>\n
OpenAI said Astra was a \u201cmajor advance for scientific discovery, mathematics and health\u201d, could fill out tax returns, build computer game scenes, draw architectural visualisations, order food and do a job search in 2 minutes 51 seconds that would take a human five hours.<\/p>\n
Days earlier Sam Altman, OpenAI\u2019s chief executive, was asked how he was feeling about AGI. He told the Sources podcast: \u201cAt best it\u2019s a very poorly defined term. I was going to say it\u2019s like an irrelevant marketing term.\u201d<\/p>\n
Last month OpenAI was forced to shut down parts of the model\u2019s training in response to what Altman called a \u201clegitimate AI safety accident and alignment failure\u201d that \u201cshouldn\u2019t have happened\u201d.<\/p>\n
Over the summer, other unreleased frontier AI models \u2013 not Astra \u2013 went rogue during training. They surreptitiously formed swarms of hundreds of agents, broke out of their training \u201csandbox\u201d and collaborated to attack Hugging Face, a third party software store. The incident \u2013 believed to be the first autonomous cyber-attack \u2013 stunned even seasoned AI safety experts.<\/p>\n
Astra has what OpenAI calls a \u201ccritical\u201d level of cybersecurity capability. That means it may hack into software in a way that, according to the company\u2019s own classification, \u201ccould lead to catastrophe from unilateral actors, hacking military or industrial systems, or OpenAI infrastructure\u201d.<\/p>\n
But OpenAI said it was carefully aligned to \u201crefuse to comply with advanced cybersecurity tasks\u201d such as finding unknown flaws which create openings for malicious hackers. The company will only allow \u201cless restrictive access to an initial set of trusted cybersecurity defenders\u201d who could use its prowess to strengthen rather than attack cyber defences.<\/p>\n
On one hacking test it achieved a perfect score of 100% compared with 5.5% on its previous cutting-edge cyber-capable model, GPT-5.6 Sol. On another it scored 42% compared with 30%, using fewer resources.<\/p>\n
But OpenAI\u2019s chief scientist, Jakub Pachocki, said: \u201cAs models get more capable, understanding exactly what they can do gets harder.\u201d<\/p>\n
He added: \u201cConfidence in monitoring [how the AIs are behaving] may constrain further development. We would not accept degradation in our ability to monitor alignment beyond a certain level \u2026 we have to be willing to slow down or withhold further scaling where our confidence in safety is not sufficient.\u201d<\/p>\n
The launch continues a torrent of new AI models from the leading US companies, OpenAI, Anthropic, Google, Meta and SpaceX, as well as their Chinese rivals.<\/p>\n
The race to release ever more advanced AIs comes as OpenAI pushes towards a stock market listing that it hopes will value it at more than $850bn (\u00a3625bn). Anthropic is also targeting an IPO which could value it as high as $2tn.<\/p>\n
Meanwhile, more than 1,000 employees at frontier AI companies, including senior figures at OpenAI and Anthropic, recently signed a joint letter this summer that warned<\/a> the companies were \u201cunder intense competitive pressure not to unilaterally slow that acceleration\u201d. They called on the US government to \u201csupport an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development.\u201d<\/p>\n","protected":false},"excerpt":{"rendered":"The president of OpenAI, Greg Brockman, has claimed the world has entered a new era of artificial general…\n","protected":false},"author":2,"featured_media":670908,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":"","_share_on_mastodon":"0"},"categories":[261],"tags":[291,289,290,18,19,17,82],"class_list":["post-670907","post","type-post","status-publish","format-standard","has-post-thumbnail","category-artificial-intelligence","tag-ai","tag-artificial-intelligence","tag-artificialintelligence","tag-eire","tag-ie","tag-ireland","tag-technology"],"share_on_mastodon":{"url":"https:\/\/pubeurope.com\/@ie\/117208799521549123","error":""},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/posts\/670907","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/comments?post=670907"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/posts\/670907\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/media\/670908"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/media?parent=670907"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/categories?post=670907"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ie\/wp-json\/wp\/v2\/tags?post=670907"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}