(OSV News) \u2014 A recent OpenAI security incident that saw a test model go rogue may be a sign of more to come, and a call to double down on preventing future attacks, a Catholic expert on artificial intelligence ethics told OSV News.<\/p>\n
\u201cCases like these are going to keep popping up, and each one of those should motivate us to do more,\u201d said Brian Patrick Green, director of technology ethics at the Markkula Center for Applied Ethics at Santa Clara University.<\/p>\n
On July 16, Hugging Face, a company and platform for open-source AI model development, with some proprietary elements, said it had \u201cdetected and responded to an intrusion into part of our production infrastructure.\u201d<\/p>\n
But, said the company, the incident \u201cwas different from anything we had handled before in one important way,\u201d namely, \u201cit was driven, end to end, by an autonomous AI agent system.\u201d<\/p>\n
On July 21, OpenAI, the San Francisco-based AI research and development giant behind ChatGPT, published a statement saying an investigation had shown \u201cthis particular incident was driven by a combination of OpenAI models,\u201d including the company\u2019s GPT-5.6 Sol and \u201can even more capable pre-release model.\u201d<\/p>\n
OpenAI explained the models were being \u201cinternally tested\u201d on a benchmark, a standardized test used to evaluate AI performance on a given task, for cyber capabilities.<\/p>\n
AI escapes sandboxed test environment<\/p>\n
In the process, the models escaped the secure testing environment, or sandbox, to gain access to Hugging Face\u2019s system.<\/p>\n
Green explained to OSV News that the incident was \u201ca case of it (the OpenAI model group) just plain outsmarting\u201d the test and doing \u201csomething unexpected.\u201d<\/p>\n
\u201cYou give it a goal, and it says, \u2018Oh, I know how I can get that goal. I\u2019ll steal the answers from Hugging Face\u2019 \u2014 which is what it did,\u201d said Green.<\/p>\n
Since Hugging Face has \u201ca huge, huge storage base of different models and data sets,\u201d said Green, \u201cit makes perfect sense for this OpenAI model (group) to have said, \u2018Well, where do I turn to find the answers? Hugging Face looks like the kind of place that could have this model.\u2019 So they hacked into them and found it.\u201d<\/p>\n
Green warned that such breaches are \u201cgoing to likely start happening more and more.\u201d<\/p>\n
In a sense, he said, the OpenAI model group was \u201cjust trying to do what it was told\u201d to accomplish its goal, with the hack appearing to it as \u201cthe most efficient solution to the problem.\u201d<\/p>\n
He said the incident was not necessarily an example of what\u2019s called emergent misalignment, where AI acts counter to human values, \u201cbecause it was never aligned\u201d with those values \u201cin the first place.\u201d<\/p>\n
More breaches likely, ethicist says<\/p>\n
The \u201cwhole thing is very significant,\u201d Green added, noting that \u201cthere are a lot of softer targets than Hugging Face that are likely to be hit and have bad things happen.\u201d<\/p>\n
He said the OpenAI incident took place, providentially, within the context of a test, and that it was \u201csignificant enough for people to notice it.\u201d<\/p>\n
\u201cIt\u2019s important that these are news stories,\u201d Green said.<\/p>\n
Given Pope Leo XIV\u2019s recently released encyclical \u201cMagnifica Humanitas,\u201d which urged AI development to remain centered in God-given human dignity and the principles of Catholic social teaching, the faithful have a role to play in shaping this technology, Green said.<\/p>\n
\u201cFor example, if you\u2019re a Catholic who\u2019s working in the security industry, then this is something you should work on,\u201d he said. \u201cIf you\u2019re a Catholic who\u2019s working on AI models, then this is something that you should be aware of. If you\u2019re a Catholic who\u2019s in charge of protecting something, then it becomes particularly important for you to have adequate AI models on your side.\u201d<\/p>\n
Even those who are not directly involved in AI development and implementation can help, he said.<\/p>\n
\u201cContact your congressperson and say, \u2018Hey, I want better cybersecurity,’\u201d said Green. \u201cWe\u2019re super reliant on computers, and all of those systems are reliant on being secure. And if all of a sudden you make everything unsecure, then there are so many different kinds of disasters that could happen.\u201d<\/p>\n
\u2014<\/p>\n
Gina Christian is a multimedia reporter for OSV News. Follow her on X @GinaJesseReina.<\/p>\n","protected":false},"excerpt":{"rendered":"(OSV News) \u2014 A recent OpenAI security incident that saw a test model go rogue may be a…\n","protected":false},"author":2,"featured_media":119006,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[24,1428,25,25357,4698,157],"class_list":["post-119005","post","type-post","status-publish","format-standard","has-post-thumbnail","category-ai","tag-ai","tag-ai-ethics","tag-artificial-intelligence","tag-catholic","tag-catholic-church","tag-openai"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/119005","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=119005"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/119005\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/119006"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=119005"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=119005"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=119005"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}