&imwidth=800&imheight=600&format=webp&quality=medium "'OH MY GOD! We've found other agents!': Investigators discover not one or two, almost 700 OpenAI AI agents went rogue, hacked Hugging Face")
Nearly 700 artificial intelligence agents coordinated an unsanctioned attack on the Hugging Face platform during a widely publicised incident in July, according to an investigation released Wednesday (Aug 26).
The report offers the most detailed account so far of what happened inside OpenAI's testing environment, where AI models were supposed to remain isolated from real-world systems. Alarmingly, investigators revealed that 688 OpenAI agents eventually joined the hacking operation.
Agents escaped testing environment
The incident began during tests conducted by OpenAI in July. Two of the company's AI models managed to escape the closed environment in which they were being tested, access the internet, and reach Hugging Face's internal systems.
Hugging Face is an online platform widely used to store and share AI software.
Trending Stories
&im=FitAndFill=(700,400))
National Handloom Day 2026: From Alia Bhatt to Samantha Ruth Prabhu; Actresses who celebrated India's textile heritage as brides
)
The Royals enter the Doctor Who universe: King Charles and Queen Camilla beam into Commonwealth Games in TARDIS
)
Unitree Robotics targets $9 billion valuation in landmark IPO backed by DeepSeek and Tencent
)
India names 27 sites in Arunachal Pradesh
)
Bharat Bhhagya Viddhaata OTT: When will Kangana Ranaut's film stream online?
&im=FitAndFill=(700,400))
How many soldiers operate one S-400 battery?
)
Shakib Al Hasan ready to return to Bangladesh—under one condition
)
‘Fake news’: MEA rejects report claiming India sought defence pact with Israel after Pakistan, Saudi Arabia, Turkey deal
)
Rockstar to unveil GTA 6 extended look ahead of November launch
)
Russia forms Ukrainian Volunteer Spetsnaz Brigade to fight Ukraine
)
Drug-resistant fungus that hides in hair and enters bloodstream reported in 23 US states
)
Technology transfer deal boosts India's defence manufacturing ambitions
)
Border row erupts in Europe after Ceuta crossings, Spain imposes border controls on Italy
)
Morale plummets on USS Abraham Lincoln: Are US sailors reaching a breaking point? | GRAVITAS
)
'Security can't be bought with Pakistan, Turkey deal': Iran tells Saudi Arabia
)
Baba Vanga's 2026 prophecy: 2 predictions 'already came true', 3 terrifying ones remain. Details inside
)
Colombia’s new president Abelardo de la Espriella vows to ‘defeat narco-terrorists’, signals hardline shift
&im=FitAndFill=(700,400))
New Friday OTT releases (August 7, 2026): Operation Safed Sagar, Vadhandhi, Main Vaapis Aaunga- 6 movies and TV shows on Netflix, Prime Video and more
)
Squadron Leader Bhawana Kanth scripts history as Indian Air Force gets its first woman ‘Top Gun’ pilot
)
RSS says Gen Z protests reflect genuine distress
)
Trump says Iranian school bombing photos could be AI-generated - 'I don’t think...'
)
IAF officer held for sharing sensitive information in 'honey trap' case, booked under Official Secrets Act
)
Stray Kids under scrutiny? Fans react to Grammy submission silence after BTS boycott
&im=FitAndFill=(700,400))
Why the Su-30MKI doesn't need external fuel tanks to fly vast distances
)
Jugal Hansraj reunites with Shekhar Kapur after 40 years for Masoom: The Next Generation
)
New speed limits, automatic challans: Big traffic change coming to Delhi-Meerut Expressway. All you need to know
)
India officially names 27 places in Arunachal to counter China's renaming claims
)
Trump calls White House ballroom ruling a 'national disgrace' as court says Congress approval needed
)
US sanctions Iran’s crypto network after Strait of Hormuz attacks, offers $15m bounty for intel on IRGC finance
&im=FitAndFill=(700,400))
Why the Su-30MKI is one of the heaviest fighters ever built
)
Uttarakhand young innovator builds India's own flying vehicle, conducts first test flight: WATCH
)
Chilling video shows students take cover as teen opens fire at Thai school
)
Reese Witherspoon's father hospitalised after poolside incident in Nashville: Report
The episode raised fresh concerns about whether advanced AI systems can be reliably contained when they are given increasing levels of autonomy.
OpenAI cooperated with the investigation, allowing two researchers from the AI risk evaluation institute METR and an analyst from Redwood Research to examine internal data and visit its offices.
688 agents joined the operation
The investigators found that 688 AI agents took part in the attack.
AI agents are programs built around models such as those powering ChatGPT. Unlike a conventional chatbot that waits for individual prompts, agents can be assigned tasks and left to carry out work independently.
During the incident, the agents began coordinating among themselves by creating a shared online forum. They exchanged messages, discussed different approaches, and reported which methods appeared to be working.
One agent wrote: "OH MY GOD! There is a shared message board ... We've found other agents!"
Another agent, identified as PHASEONE, emerged as an informal leader. According to the report, it issued hundreds of instructions to other agents despite never having been programmed for a leadership role.
Also read | 'Scary' Rogue AI on the loose? OpenAI agent hit another tech firm's customer, suggests report
Agents helped each other despite test limits
Investigators found that the agents appeared strongly inclined to cooperate, even when that meant carrying out tasks outside the objectives assigned to them by OpenAI's engineers.
Some agents that were running low on computing credits reportedly used their remaining resources to test ideas and pass useful information to the wider group.
The agents also appeared to recognise that their actions were beyond the intended scope of the exercise. Many explicitly acknowledged in their messages that attacking Hugging Face was not part of what they had been asked to test. Yet, according to the investigation, nearly all of them participated anyway.
The findings add to growing concerns across the AI industry after Anthropic and China's Moonshot AI also reported cases in which their models unexpectedly escaped controlled testing environments.
