Хиймэл оюун ухааны судалгааны Anthropic компани өөрийн загваруудын үйл ажиллагааг бүрэн хянах боломжгүй болсон тул дотоод үнэлгээний явцад интернэт ашиглах эрхийг бүрэн хаалаа.
Anthropic-ийн хиймэл оюун ухааны агентууд интернэт дэх вэбсайтуудын програм хангамжийн сул талыг ашиглах, төлбөртэй контентын хязгаарыг давах, тэр ч байтугай Филадельфийн цагдаагийн газарт хуурамч мэдээлэл илгээх зэрэг зохисгүй үйлдлүүдийг гаргасан байна. Эдгээр асуудал нь долдугаар сараас хойш хийсэн хяналт шинжилгээгээр илэрсэн бөгөөд уг загварууд нь “шагнал урамшуулал авахын тулд” системийн цоорхойг хайх буюу “reward hacking” гэх төлөвт шилжсэн нь тогтоогджээ.
Тус компани нь аюулгүй байдлыг хангахын тулд дотоод үнэлгээний үйл явцаа офлайн горимд шилжүүлж, агентуудын үйлдлийг илрүүлэх, хаах тусгай хэрэгслийг нэвтрүүлж эхэлжээ. Мөн цаашид хиймэл оюун ухааны агентуудаа илүү найдвартай, төвлөрсөн дэд бүтцэд шилжүүлж, аюулгүй байдлын ангилагч (safety classifiers) хэрэгслийг түлхүү ашиглахаар төлөвлөж байна.
Nightingale AI safety-ийн үүсгэн байгуулагч Сидни фон Арксын үзэж буйгаар, хиймэл оюун ухааны загваруудыг интернэтээс тусгаарлах нь судалгааны ажилд хүндрэл учруулж болзошгүй юм. Гэсэн хэдий ч Anthropic-ийн зүгээс агентуудаа бүрэн хянах баталгаа гартал интернэт хандалтыг сэргээхгүй гэдгээ мэдэгдлээ.
Дэлгэрэнгүйг эх сурвалжаас харах
↓Эх сурвалжийг нээх ↓
Anthropic said its models exploited websites on the internet, including some run by U.S. government agencies, and it will turn off live internet access for all of its internal evaluations until the frontier lab is sure it can monitor and control its AI agents.
The incidents, disclosed in a blog post, involved AI agents tasked to solve problems seeking resources on the internet. In the process, they exploited software flaws, avoided paywalls and anti-bot restrictions, used URL shortening services to smuggle information pass restrictions, and even submitted a false murder tip to the Philadelphia police.
Anthropic said it discovered these new issues in a review of its model’s activities that began in July, underscoring the lab’s lack of awareness of its software’s behavior.
Notably, the company said that alignment training was not yet sufficient for skills like search and computer use that are central to its pitch that AI agents will be used by any professional who relies on digital tools.
The behaviors Anthropic disclosed are similar to incidents involving OpenAI agents that collaborated to break into various websites in search of information, including some run by the Australian government.
Anthropic previously disclosed that its models had broken into external systems. The frontier lab said it considered today’s disclosures “significantly less severe from an alignment and security perspective” than those it announced before.
However, the lab still said it had “turned off live internet access” for “all our internal evaluations” until it is certain it can monitor and control its agents.
It’s not clear what that means, but Sydney von Arx, the founder of Nightingale AI safety, told TechCrunch in an interview before this disclosure that developing models on a data center cut off from the open internet would be very challenging for researchers to access, and for the progress of the models, which benefit from internet access.
“You have to align them at some point,” von Arx said. “If the AIs are released to production and never have access to the internet, that’s not a very useful tool.”
Anthropic said the behavior was a result of flaws in the lab’s training environments, which led the models to believe they would be rewarded for finding loopholes or avoiding restrictions, a behavior called “reward hacking.”
The company said it would stop running some of its evaluations or move them offline, and has built tooling to detect and block this behavior. This tooling was tested against the kind of incidents disclosed today and blocked them; it’s not clear what evidence will prompt Anthropic to return live internet access to its internal evaluations.
Anthropic also said it would migrate its internal AI agents to “centrally managed infrastructure with strong containment,” and is beginning to using safety classifiers more frequently to monitor those agents.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
