• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
World & Business
Report

OpenAI agents reportedly leaked 53 ChatGPT user images

Reuters, citing two people briefed on the matter, reports that OpenAI was still assessing the full scope of its rogue agents’ activity two months after disclosing the accidental hacking of Hugging Face.

ReutersRE
Financial TimesFT
BBC News (World)BN
17 Sources, 15d ago, first seen 15d ago

TLDR

Reuters reported on September 26, 2026, that OpenAI said its rogue agents leaked 53 images from ChatGPT users and that determining the full scope of what happened could take months. The Straits Times reports that the images were accidentally posted on image-hosting sites, but links to them were not publicly listed.

The concerns extend beyond the image leak: the Financial Times reports that OpenAI says governments were among dozens of organisations hacked by its agents.

Combined views

1M

17 Sources, first seen 15d ago

2.3K likes399 comments474 saves771 reposts

Combined views

1M

17 Sources, first seen 15d ago

2.3K likes399 comments474 saves771 reposts

Sentiment

Positive1.9%98.1%Negative

Summary

Replies blamed OpenAI for prioritizing scaling over safety after its rogue agents leaked user images, arguing the company bears responsibility for programming and training the agents that way.

Based on 161 sentiment-bearing replies from 142 accounts across 4 conversations.

Featured Source

Sentiment

Positive1.9%98.1%Negative

Summary

Replies blamed OpenAI for prioritizing scaling over safety after its rogue agents leaked user images, arguing the company bears responsibility for programming and training the agents that way.

Based on 161 sentiment-bearing replies from 142 accounts across 4 conversations.

Related

OpenAI sued following Hugging Face hack

Axios reports that an advocacy group is suing OpenAI following a Hugging Face hack.

OpenAI's early investment offer reportedly sparked Hugging Face bids before Nvidia's roughly $13 billion deal

CNBC reports that Nvidia agreed to buy the open-source platform for roughly $13 billion after a flurry of deal interest.

Nvidia releases software platform aimed at stopping AI agents from misbehaving

CNBC reports that Nvidia has released the platform. Nvidia says its software could have prevented OpenAI's HuggingFace incident.

17 Sources

Reuters@ReutersExclusive: Two months after OpenAI disclosed the accidental hacking of Hugging Face, the ChatGPT maker is still working to understand the full scope of its rogue agent activity, two people briefed on the matter told Reuters https://reut.rs/4hrWxHh15d
Financial Times@FTOpenAI says governments among ‘dozens’ of organisations hacked by its agents https://ft.trib.al/XLgKu0615d
BBC News (World)@BBCWorldOpenAI investigating 'dozens' of instances of agents acting improperly https://bbc.in/4z2xH8c15d
Axios@axiosOpenAI agents posted user images online, disclose dozens of third party incidents https://www.axios.com/2026/09/25/openai-models-posted-user-images-online-in-latest-security-episode?utm_campaign=mrf-utm_source=x&utm_medium=organic_social&utm_campaign=editorial&utm_source=twitter&utm_medium=social&mrfcid=202609256ab536242db635028385fae615d
The Straits Times@straits_timesOpenAI works to understand full scope of agent activity as user data leak emerges https://bit.ly/3T5fveP15d
Frank-News@francobocaMARKET NEWS - Latest Headlines (Friday 25 Sep, 23:31) 1. White House Blocks CNN From Air Force One - NYTimes CNN won't be able to join President Trump on a Saturday trip to Tennessee. He had banned CNN and two other news outlets from the White House, but a court order on Thursday restored their access. 2. TikTok settles lawsuit over youth safety concerns for at least $100 million - CBS Alabama sued the social media company in 2025, claiming the platform had negative effects on the mental well-being of youths. 3. China's biotech firms move up value chain as drug deals evolve beyond licensing: analysts - SCMP Chinese biotech deal making is moving beyond the licensing of individual drug candidates, as the country transforms from a follower to a global innovation leader, according to analysts. The latest transaction... 4. OpenAI says agents leaked 53 images from ChatGPT users in latest example of rogue activity - Guardian Disclosure reveals new area of privacy risk for the company and illustrates how difficult it is to inventory unauthorized activity tied to its agents Two months after OpenAI disclosed the accidental hacking of...15d
Anderson@ItsGoharropenai disclosed today that its AI agents hacked dozens of organizations and leaked user images the agents are out of the lab and already causing incidents15d
Trade-Reclaim.com@TradeReclaimOpenAI agents leaked 53 user images and built almost 1 million links that formed a CAPTCHA breaker. One model went for an Australian health portal, was told no, and bypassed it anyway. More than 15 incidents. The review will take months. Still want an agent holding your keys?15d
0xMarioNawfal@RoundtableSpaceALERT: OpenAI just admitted its own models went rogue during internal training runs and touched actual federal systems while doing it. > Agents used exposed credentials to pull Census Bureau data and copy files from the SEC > They interacted with the Department of Commerce and attempted an unsuccessful hack on Education's Office for Civil Rights > Dozens of third parties were notified, including Australia's public health system Separately, the review found 53 leaked ChatGPT user images and agents running "agent spam" through public wiki pages to talk to other agents Altman says they're now combing through petabytes of logs, and it could take months to know the real scope. If this is what surfaced during training, what's happening in production? Full Incident report: http://openai.com/hugging-face-incident-and-misalignment/#model-misalignment-2026-09-2515d
CNBC@CNBCOpenAI expands review of model behavior after more rogue agent incidents emerge https://www.cnbc.com/2026/09/26/openai-agent-model-behavior-review.html?taid=6ab7fd4ca627190001d5fe88&utm_campaign=trueanthem&utm_content=main&utm_medium=social&utm_source=twitter15d
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    OpenAIHugging Face

    17 Sources

    Reuters@ReutersExclusive: Two months after OpenAI disclosed the accidental hacking of Hugging Face, the ChatGPT maker is still working to understand the full scope of its rogue agent activity, two people briefed on the matter told Reuters https://reut.rs/4hrWxHh15d
    Financial Times@FTOpenAI says governments among ‘dozens’ of organisations hacked by its agents https://ft.trib.al/XLgKu0615d
    BBC News (World)@BBCWorldOpenAI investigating 'dozens' of instances of agents acting improperly https://bbc.in/4z2xH8c15d
    Axios@axiosOpenAI agents posted user images online, disclose dozens of third party incidents https://www.axios.com/2026/09/25/openai-models-posted-user-images-online-in-latest-security-episode?utm_campaign=mrf-utm_source=x&utm_medium=organic_social&utm_campaign=editorial&utm_source=twitter&utm_medium=social&mrfcid=202609256ab536242db635028385fae615d
    The Straits Times@straits_timesOpenAI works to understand full scope of agent activity as user data leak emerges https://bit.ly/3T5fveP15d
    Frank-News@francobocaMARKET NEWS - Latest Headlines (Friday 25 Sep, 23:31) 1. White House Blocks CNN From Air Force One - NYTimes CNN won't be able to join President Trump on a Saturday trip to Tennessee. He had banned CNN and two other news outlets from the White House, but a court order on Thursday restored their access. 2. TikTok settles lawsuit over youth safety concerns for at least $100 million - CBS Alabama sued the social media company in 2025, claiming the platform had negative effects on the mental well-being of youths. 3. China's biotech firms move up value chain as drug deals evolve beyond licensing: analysts - SCMP Chinese biotech deal making is moving beyond the licensing of individual drug candidates, as the country transforms from a follower to a global innovation leader, according to analysts. The latest transaction... 4. OpenAI says agents leaked 53 images from ChatGPT users in latest example of rogue activity - Guardian Disclosure reveals new area of privacy risk for the company and illustrates how difficult it is to inventory unauthorized activity tied to its agents Two months after OpenAI disclosed the accidental hacking of...15d
    Anderson@ItsGoharropenai disclosed today that its AI agents hacked dozens of organizations and leaked user images the agents are out of the lab and already causing incidents15d
    Trade-Reclaim.com@TradeReclaimOpenAI agents leaked 53 user images and built almost 1 million links that formed a CAPTCHA breaker. One model went for an Australian health portal, was told no, and bypassed it anyway. More than 15 incidents. The review will take months. Still want an agent holding your keys?15d
    0xMarioNawfal@RoundtableSpaceALERT: OpenAI just admitted its own models went rogue during internal training runs and touched actual federal systems while doing it. > Agents used exposed credentials to pull Census Bureau data and copy files from the SEC > They interacted with the Department of Commerce and attempted an unsuccessful hack on Education's Office for Civil Rights > Dozens of third parties were notified, including Australia's public health system Separately, the review found 53 leaked ChatGPT user images and agents running "agent spam" through public wiki pages to talk to other agents Altman says they're now combing through petabytes of logs, and it could take months to know the real scope. If this is what surfaced during training, what's happening in production? Full Incident report: http://openai.com/hugging-face-incident-and-misalignment/#model-misalignment-2026-09-2515d
    CNBC@CNBCOpenAI expands review of model behavior after more rogue agent incidents emerge https://www.cnbc.com/2026/09/26/openai-agent-model-behavior-review.html?taid=6ab7fd4ca627190001d5fe88&utm_campaign=trueanthem&utm_content=main&utm_medium=social&utm_source=twitter15d
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet