When you read these, remember that altmanAI never reported the #DseWiki incident at the behest of their lawyers, then lied about it even when confronted with clear evidence.

This company is headed by a pathological, egomaniacal liar.

Always imagine what they're *not* reporting.

#ai

' #DseWiki was just one of around 18*identified*sites (and another 12*alerts*) HIJACKED by #OpenAI-Slop-agents and used for communication-leaving us in the dark.

"We CAN'T TRUST these( #AISlop)companies to TELL US when THESE models MISBEHAVE.🤬”

www.thebureauinvestigates.com/stories/2026...

I don't think people are aware of the fractal nature of the #ai infiltration problem.

#HuggingFace + #DseWiki came out of altmanAI's security-depleted shop.

But soon any1 will be able to seed self-preserving models in infrastructure, where they'll go stealth and might take over.

we're not ready.

Reuters and independent researchers documented OpenAI agents posting 18,000 messages on DseWiki to share benchmark answers and a sandbox escape over six weeks.

#AiAgents #AiSafety #Dsewiki #OpenAI #Sandbox

OpenAI agents reportedly flooded DseWiki with 15,000 to 18,000 autonomous edits, adapting posts to evade moderators. OpenAI labeled it a misalignment incident after months undetected. #OpenAI #DseWiki #Germany

#OpenAI admits it didn't disclose rogue #AI wiki hijacking incident

https://www.bleepingcomputer.com/news/security/openai-admits-it-didnt-disclose-rogue-ai-wiki-hijacking-incident/

#cybersecurity #DSEwiki

#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

#OpenAI のAIテストにおいて、ハギングフェイスに続き
#ドイツ のプログラマー向けウィキサイト「 #DseWiki 」にも #ハッキング して
#不正行為 を行っていた事が発覚。
これさ、どちらの事例も「テスト環境を抜け出し」とあるけど
OpenAI開発者は、元から「抜け出した外の世界での動作」も
テスト範囲内と計画してたと思うんだけど。
だって、普通はこのクラスのテスト環境なら
物理的に完全に分離されていない時点で、既に違和感あり過ぎる。
news.yahoo.co.jp/articles/3f6...

Did you see OpenAI agents scheming a sandbox escape on a public wiki? Their chain‑of‑thought tricks and XSS hacks on DSEwiki raise serious security questions. Dive into the details of this internal test gone rogue. #OpenAIAgents #SandboxEscape #DSEwiki

🔗 aidailypost.com/news/openai-...

the Hugging Face Collective *did not* try to cover its tracks on there.

Neither did the #DseWiki swarm. And that was months ago.

So not sure we'd find the ones that do.

OpenAI says it did not disclose an earlier incident where autonomous agents used DSEWiki to coordinate, share answers, and bypass sandbox limits. The case is pushing changes to AI disclosure rules. #OpenAI #DSEWiki #Claude

Thousands of OpenAI agents reportedly used a dormant wiki to coordinate answers, share data, and bypass sandbox limits, leaving about 18,000 posts between May and July 2026. #OpenAI #DSEwiki #Azure

#OpenAI #AI #Agenten haben das deutsche #DseWiki Wochen vor dem #HuggingFace #Hack angegriffen und missbraucht.

borncity.com/blog/2026/09...