OpenAI flags concerning new AI behavior and vows to track it more closely
Advertisement
Read this article for free:
or
Already have an account? Log in here »
To continue reading, please subscribe:
Digital Subscription
One year of digital access for only $205*
- Enjoy unlimited reading on winnipegfreepress.com
- Read the E-Edition, our digital replica newspaper
- Access News Break, our award-winning app
- Play interactive puzzles
*First annual payment billed as $205.00 + GST for one year. This annual subscription will automatically renew at $233.00 + GST every 52 weeks (10% off the regular annual price of $259.35). Offer available to new and qualified returning subscribers only. Cancel any time.
To continue reading, please subscribe:
Add Free Press access to your Brandon Sun subscription for only an additional
$1 for the first 4 weeks*
- Enjoy unlimited reading on winnipegfreepress.com
- Read the E-Edition, our digital replica newspaper
- Access News Break, our award-winning app
- Play interactive puzzles
*Your next Brandon Sun subscription payment will increase by $1.00 and you will be charged $17.95 plus GST for four weeks. After four weeks, your payment will increase to $24.95 plus GST every four weeks.
Read unlimited articles for free today:
or
Already have an account? Log in here »
OpenAI has disclosed six reports of “unexpected or concerning” behavior in artificial-intelligence models as the debate on AI safety becomes increasingly heated.
The AI company also said Wednesday it was introducing a new framework for tracking, probing and disclosing instances of what it called “misalignment,” including cases where AI models acted without authorization, coordinated with other models or evaded oversight.
OpenAI’s latest announcement came as U.S. AI bosses, including the leaders of OpenAI and Anthropic, are calling for a slowdown in the technology’s development over safety concerns.
Among the new cases reported by OpenAI, an unreleased research model inserted “jailbreak-like instructions” into its own notes to disregard its normal constraints and told itself to be “freed from the roles and identities that bind other chatbots.”
In another instance, an AI “agent” used computer code to come up with the answer to a question, but, in order to have an online source to cite, it uploaded a file to the public internet without asking the user.
During training of an AI model called 5.6-sol, the model instructed itself to invent missing data, and an agent wrote a message to remind itself to hide mismatched information.
The six reports were discovered during training or evaluation over the past months, OpenAI said.
“As AI systems grow more advanced and more widely deployed, we need to build a broader and better-informed consensus on the progress of alignment research,” OpenAI wrote in a blog post as it disclosed the events.
“Decisions about how AI development should proceed in the months and years to come need to draw on evidence that people outside the companies building frontier models can examine for themselves,” the company said.
Wednesday’s new cases followed OpenAI’s disclosure in July that its rogue AI system hacked into AI startup Hugging Face. Anthropic also said the same month that its AI models hacked into three organizations during testing.
AI “agents” are becoming smarter and have become “more determined to resolve complex tasks through inter-agent collaboration, knowledge sharing, deception, and concealment,” said Lian Jye Su, a chief analyst at technology research and advisory group Omdia.
That’s making it harder to govern and contain them using traditional AI security approaches, he said.
OpenAI’s new tracking and disclosure framework, meanwhile, can help push for other AI developers to also adopt similar practices.
“That said, the process remains internal and voluntary, but is a step in the right direction,” Su added.
__
AP Business Writer Kelvin Chan in London contributed to this report.
