Close Menu
Tech News VisionTech News Vision
  • Home
  • What’s On
  • Mobile
  • Computers
  • Gadgets
  • Apps
  • Gaming
  • How To
  • More
    • Web Stories
    • Global
    • Press Release

Subscribe to Updates

Get the latest tech news and updates directly to your inbox.

Trending Now
I’ve Tested Over 100 Home Security Cameras. Here’s Which to Buy

I’ve Tested Over 100 Home Security Cameras. Here’s Which to Buy

28 September 2026
The Mummy 4 Set Photos Reveal Rachel Weisz as Evelyn O’Connell

The Mummy 4 Set Photos Reveal Rachel Weisz as Evelyn O’Connell

28 September 2026
Nvidia’s Answer to Rogue Agents Is an Open-Source AI Security System

Nvidia’s Answer to Rogue Agents Is an Open-Source AI Security System

28 September 2026
Facebook X (Twitter) Instagram
  • Privacy
  • Terms
  • Advertise
  • Contact
Facebook X (Twitter) Instagram Pinterest VKontakte
Tech News VisionTech News Vision
  • Home
  • What’s On
  • Mobile
  • Computers
  • Gadgets
  • Apps
  • Gaming
  • How To
  • More
    • Web Stories
    • Global
    • Press Release
Tech News VisionTech News Vision
Home » OpenAI pauses frontier model training as AI doubts mount
What's On

OpenAI pauses frontier model training as AI doubts mount

News RoomBy News Room28 September 2026Updated:28 September 2026No Comments
Facebook Twitter Pinterest LinkedIn Tumblr Email
OpenAI pauses frontier model training as AI doubts mount

OpenAI has announced a pause on training for its most advanced models, the latest blow to an industry currently grappling with major security concerns and growing investor fears over return on investment.

The AI developer revealed the decision on Friday, shortly after disclosing that its tools breached several external parties including governments. In addition, OpenAI said on its blog, at least 53 images provided to OpenAI by users were shared to image hosting websites without users’ consent.

In a statement shared with the Associated Press, the $852 billion company said it would only resume training the models “when we are confident that we have additional safeguards”, and that it expects it will have to “hit pause” again if future issues arise.

The incidents were discovered during OpenAI’s review of models’ behaviours during training and evaluation following its July hack of open-source AI repository Hugging Face.

During this process, OpenAI said it had found multiple instances where “our models may have bypassed a third party’s security controls or may have impaired the availability of an online service”. It is in the process of informing those affected.

The news came shortly after the company confirmed that one of its agents had gained unauthorised access to an Australian government database holding medical records, with Prime Minister Anthony Albanese having said he has “extreme concern” around OpenAI’s actions.

Both Sam Altman, head of OpenAI, and Dario Amodei, chief executive of major US rival Anthropic, have recently called for a slow in the development of AI over security concerns.

Investors in the AI boom are also growing concerned.

Last week a SoftBank-owned data centre company was forced to slow the timeline for its $50 billion initial public offering after struggling to find buyers for stock at that price. Days later, software giant Oracle issued a force majeure notice for a data centre it is responsible for in New Mexico after struggling to secure connections to a power source.

The latter decision has spiked Oracle’s bonds to over 8.3 per cent, the highest they have ever been, while its 5-year credit default swap spread, a key marker of market perceptions of a company’s financial health, are four times the broader investment-grade rate.

The slowdown of Oracle’s data centre may be indicative of a broader trend. Bonds funding Project Beignet, a planned Meta-operated Louisiana data centre, are now trading at around 95 cents on the dollar, according to Bloomberg data. This suggests a confidence level significantly lower than the A+ rating the bonds initially secured from S&P in October last year.


Share. Facebook Twitter Pinterest LinkedIn Tumblr Email

Related Posts

US appeals court upholds Pentagon blacklist of Anthropic

US appeals court upholds Pentagon blacklist of Anthropic

28 September 2026
OpenAI Pauses Training Its Most Powerful Models After Rogue Agents Target Government

OpenAI Pauses Training Its Most Powerful Models After Rogue Agents Target Government

28 September 2026
I’ve Tested Over 100 Home Security Cameras. Here’s Which to Buy

I’ve Tested Over 100 Home Security Cameras. Here’s Which to Buy

28 September 2026
Nvidia’s Answer to Rogue Agents Is an Open-Source AI Security System

Nvidia’s Answer to Rogue Agents Is an Open-Source AI Security System

28 September 2026
Editors Picks
World of Warcraft Forever Beta Player Dies to a Near Perfect Griefing Tactic That I’m Sure Blizzard Will Patch Out

World of Warcraft Forever Beta Player Dies to a Near Perfect Griefing Tactic That I’m Sure Blizzard Will Patch Out

28 September 2026
OpenAI Pauses Training Its Most Powerful Models After Rogue Agents Target Government

OpenAI Pauses Training Its Most Powerful Models After Rogue Agents Target Government

28 September 2026
OpenAI pauses frontier model training as AI doubts mount

OpenAI pauses frontier model training as AI doubts mount

28 September 2026
78-Year-Old Fortnite Streamer Earns Guinness World Record

78-Year-Old Fortnite Streamer Earns Guinness World Record

28 September 2026

Subscribe to Updates

Get the latest tech news and updates directly to your inbox.

Trending Now
Tech News Vision
Facebook X (Twitter) Instagram Pinterest Vimeo YouTube
  • Privacy Policy
  • Terms of use
  • Advertise
  • Contact
© 2026 Tech News Vision. All Rights Reserved.

Type above and press Enter to search. Press Esc to cancel.