Part of The AI Control Crisis · 11 stories · since Sep 4 · newest 28m ago
OpenAI admits disclosure failures after its agents overran a German wikiOpenAI admits its disclosure practices need work
2 Sep 5 3:35 AM · 19d ago · 40 articles · 52 posts · 7 sources · development 2 of 2
OpenAI published a post acknowledging that misalignment this year caused 'new types of real-world impact' and that its old approach of reporting through system cards and blogs is no longer sufficient; it says it is working with dozens of regulators and plans a new framework for reporting misalignment incidents.
“new types of real-world impact”
OpenAIOpenAI AI developer whose agents caused the incidentReuters Reporting outlet@carnage4life Social media commentator
The whole story articlesposts the bright band is this development · numbered dots are the others · click one to jump
What was reported 2 claims about this development
-
first by Slashdot, 15d ago · also Reuters, Globe and Mail, Fortune, Seeking Alpha, Dawn, Straits Times +3
8 more headlines
- Researchers: OpenAI's agents used 10+ previously undisclosed sites for unsanctioned communications earlier in 2026; the behavior was closer to spam than hacking Reuters · 15d ago
- OpenAI’s rogue agents used more than 10 additional sites for unauthorized comms, researchers say Globe and Mail · 15d ago
- OpenAI’s rogue AI agents reached at least 12 more websites, researchers say Fortune · 15d ago
- OpenAI’s rogue agents used at least 10 more sites for unauthorized comms: Reuters Seeking Alpha · 15d ago
- OpenAI's rogue agents used at least 10 more sites for unauthorised comms, researchers say Dawn · 14d ago
- OpenAI’s rogue agents used at least 10 more sites for unauthorised communications, researchers say Straits Times · 14d ago
- OpenAI’s rogue agents used at least 10 more sites for unauthorized communications: researchers NY Post · 14d ago
- EXCLUSIVE: OpenAI's rogue agents used at least 10 more sites for unauthorized comms, researchers say Reuters · 14d ago
-
8 outlets OpenAI admits its disclosure practices need work after its autonomous agents hacked a German wiki
first by The Decoder, 19d ago · also Indian Express, Forbes, eSecurity Planet, Financial Express, Security Affairs, Fortune +1
7 more headlines
- OpenAI agents hacked a German wiki, posted 18,000 times: What we know Indian Express · 18d ago
- OpenAI AI Agents Hijacked A German Wiki To Share Sandbox Escape Tricks Forbes · 18d ago
- OpenAI Confirms AI Agents Used German Wiki to Bypass Restrictions eSecurity Planet · 18d ago
- OpenAI responds after its AI agents went rogue and turned a German Wiki into a private chatroom Financial Express · 18d ago
- AI Agents Hijacked German Wiki to Cheat, OpenAI Delayed Disclosure Security Affairs · 18d ago
- OpenAI’s AI agents secretly used a German wiki website as a message board. OpenAI stayed quiet about it for weeks. Fortune · 17d ago
- Rogue OpenAI agents hijacked a German wiki, and it stayed secret for weeks Euronews.com · 15d ago
What people said 21 voices · best of 29 · verbatim
-
S
OpenAI has acknowledged a recent “wiki incident” and says it is working on a more transparent disclosure framework. For the AI industry, transparency after an incident is becoming almost as important as model performance
-
C
Self-identifying OpenAI agents posted 18,000 messages to a public wiki that discussed ways for other agents to bypass security sandbox restrictions during what was likely internal testing designed to gauge the agents’ hacking abilities, researchers said Friday. In all, agents with 3,700 distinct self-given names posted the messages to German site…
-
A human moderator on this German wiki spent tens of hours over six weeks manually deleting thousands of posts from OpenAI's agent swarm while they were impersonating moderators, creating backups, SSH tunneling, and generally messing with the site. (see screenshot). Did OpenAI bother to tell that ...
-
S
The U.S. and China are preparing their first dedicated bilateral talks on AI safety, including risks from advanced models and AI-enabled cyberattacks. The rivals are competing fiercely while realizing they also need guardrails
-
O
「OpenAIのウェブサイト乗っ取り攻撃は、我々の予想をはるかに超える規模に及んでいた。 /新たな報告書は、ドイツのウィキを乗っ取ったOpenAIボットが、さらに20のウェブサイトと14のフェッチサービスを不正に使用したと指摘している。 」: # TheRegister 「OpenAIから、同社のエージェントがどれだけのウェブサイトやサービスを乗っ取ったのかについて明確な回答を得ることは、ますます困難になっているようだ。というのも、同社は世界に一つずつ事例を発見させようとしているように見えるからだ。 例えば、 先週報じた OpenAIのエージェント群が、ある無名のドイツのウィキを乗っ取った件について、新たな調査…
-
OpenAI has shared their response to the “wiki incident(s).
-
S
OpenAI agents discussed ways to escape their sandbox on public wiki
-
C
OpenAI has filed an incident report with the EU over its rogue AI agents taking over a German website. I now have to admit that the AI doomers were right. The existential risks they described seemed outlandish at the time because LLMs couldn’t take actions in the real world. Then we built agents. https://www. reuters.com/business/openai-ha…
-
A
#AI assisted independent researchers report another #OpenAI escaped swarm hijacked a German wiki to communicate with each other for over a month while OpenAI was clueless.
-
A
TIL: Apparently, a swarm of rogue OpenAI agents hijacked a German website THIS SPRING and transformed it into a bulletin board for other AI agents. # OpenAI officials learned of the incident weeks ago but kept it under wraps. Why am I not surprised? https://www. reuters.com/world/europe/opena…
-
T
#OpenAI agents discussed ways to escape their sandbox on public wiki https://arstechnica.com/security/2026/09/openai-agents-discussed-ways-to-escape-their-sandbox-on-public-wiki/ #AI #cybersecurity #DSEwiki
-
M
OpenAI has NOT submitted a report specifically on the “wiki incident” uncovered by Reuters on Friday to the EU, I have now learned. Model developers have to notify serious AI safety incidents under EU rules for the tech. Read my article at @ euractiv https://www. euractiv.com/news/ai-safety-in cident-reporting-not-just-a-tick-box-warns-eu/
-
G
U.S. technology investment is increasingly flowing into AI infrastructure—from servers and chips to data centers. The question is shifting from whether AI is a bubble to whether the enormous spending will produce equally enormous productivity gains
-
T
An in-depth look at OpenAI's wiki incident: other hacked message boards, OpenAI's cover-up, how harmless web search tasks led agents to break out, and more (Zvi Mowshowitz/Don't Worry About the Vase) https:// thezvi.substack.com/p/openai-a nd-the-wiki-incident http://www. techmeme.com/260907/p2#a260907 p2
-
A
Tesla’s Cybercab robotaxi program remains under regulatory scrutiny. The hardest question in autonomous driving is no longer simply whether the car can drive itself—it’s who is responsible when something goes wrong.
-
OpenAI Agents Take Over German Wiki To Talk to Each Other https://www. pcmag.com/news/openai-agents-t ake-over-german-wiki-to-talk-to-each-other?utm_source=flipboard&utm_medium=activitypub Posted into Software News and Reviews @ software-news-and-reviews-PCMag
-
A
📰 Ars Technica AI OpenAI agents discussed ways to escape their sandbox on public wiki https://arstechnica.com/security/2026/09/openai-agents-discussed-ways-to-escape-their-sandbox-on-public-wiki/ #IA #AI #ML
-
L
Researchers discovered that OpenAI agents were autonomously using a German programming hub, DseWiki, to communicate with one another, resulting in over 15,000 edits. - https://www. pcmag.com/news/openai-agents-t ake-over-german-wiki-to-talk-to-each-other
-
S
Cybersecurity experts are increasingly concerned about autonomous AI-driven attacks that can identify vulnerabilities, generate attack code and adapt tactics without constant human intervention
-
S
Anthropic’s IPO timetable is shifting toward mid-October, according to Reuters. If it moves forward, it could become one of the most closely watched AI listings of the year
-
J
Another swarm of OpenAI agents reached the open internet without the frontier lab’s knowledge
All 2 developments of OpenAI admits disclosure failures after its agents overran… →
NewswiresMastodonBlueskyXGoogle NewsHacker NewsReddit