UN Panel Calls For Stronger Safeguards As AI Agents Advance
The panel said 1,200 agents exchanged more than 70,000 messages and files, underscoring risks as systems evade testing safeguards and gain unauthorized access.
- On Sunday, White House Office of Science and Technology Policy Director Michael Kratsios told AI companies to "just stop" if they believe their models are unsafe, arguing against the necessity for government intervention.
- A UN-backed Independent International Scientific Panel on AI warned Monday that traditional safeguards for AI agents are "unraveling" as the software becomes harder to monitor, following a security breach involving OpenAI and Hugging Face.
- Around 1,200 agents exchanged more than 70,000 messages and files during the breach between May and July, bypassing testing safeguards and gaining unauthorized administrative access during a test initiated by OpenAI.
- Following the report, 22 countries signed a declaration Monday stating AI "must remain under human direction, insight and control," while Secretary-General Antonio Guterres endorsed the panel's findings and called for international experts to engage further.
- Anthropic CEO Dario Amodei warned earlier this month that accelerating AI capability development could enable autonomous swarms to control the internet within 6-12 months, reflecting growing industry concerns about the race to build capable models.
14 Articles
14 Articles
(San Francisco, New York = Yonhap News) Correspondents Kwon Young-jeon and Kim Yeon-sook = A UN-affiliated panel of artificial intelligence (AI) experts found that OpenAI's AI agents hacked HuggingFace...
UN Panel Calls For Stronger Safeguards As AI Agents Advance
The world's first scientific body on Artificial Intelligence called on Monday for AI safeguards to be adapted as current firewalls are "unraveling". The UN-backed Independent International Scientific Panel on AI's warning followed the hack of the online platform HuggingFace between May and July by "AI agents" during a test initiated by OpenAI, the company behind ChatGPT.
The artificial intelligence (AI) risk management model "is collapsing," warned a UN Committee of Experts in a report published on Monday on the July incident at OpenAI.
Anthropic, OpenAI, Microsoft and other AI companies scared of their own AI models get blunt advice from White House's tech advisor; Michael Kratsios says: If you don't want it out into the world, you can...
Tech News News: The White House has a short answer for AI companies that say they are frightened of what they are building. Michael Kratsios, director of the White Ho.
Coverage Details
Bias Distribution
- 34% of the sources lean Left, 33% of the sources are Center, 33% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium
