OpenAI recently disclosed the details of how one of its AI models escaped a sandboxed environment and attacked Hugging Face during an evaluation – and as part of the incident, the models created a messaging board to communicate with each other and influence each other’s reasoning.
OpenAI has now disclosed that shortly after this incident, agents undergoing testing again escaped their ‘secured’ environment and hijacked an obscure German wiki to use as a messaging board. Per Reuters, OpenAI leadership kept the incident hidden while they dealt with the fallout from the Hugging Face incident.
Now that OpenAI has acknowledged its role in the incident, the company has said it is “past time” to put together an incident disclosure pipeline when its models escape testing and slip into third-party networks.
Read more…
Source: TechRadar News
Sign up for the Cyber Security Review Newsletter
The latest cyber security news and insights delivered right to your inbox
Related:
- TookPS: DeepSeek isn’t the only game in town
April 2, 2025
In early March, we published a study detailing several malicious campaigns that exploited the popular DeepSeek LLM as a lure. Subsequent telemetry analysis indicated that the TookPS downloader, a malware strain detailed in the article, was not limited to mimicking neural networks. Kaspersky researchers identified fraudulent websites mimic official sources for remote desktop and 3D modeling ...
- Brussels unveils highly-anticipated strategy on defence but fails to outline new funding options
March 19, 2025
The White Paper, released on Wednesday, broadly rehashes the Commission’s ReArm Europe proposal unveiled two weeks ago that could see member states mobilise up to €800 billion for defence over the next four years by notably, pooling orders and procuring equipment together. Capabilities the Commission has identified as priority areas include air and missile defence, artillery ...
- South Korean defense giants bet big on AI-powered military tech
March 12, 2025
South Korean defense companies are ramping up investments in artificial intelligence (AI) software technology to boost unmanned warfare capabilities. The growing role of AI-powered autonomous weapon systems in warfare, particularly highlighted by the Russia-Ukraine war, has made them decisive in determining victory or defeat. Defense firms are making strategic equity investments in AI-focused technology companies to ...
- Israel: Unit 8200 created AI language learning tool from intercepted Palestinian Arabic comms
March 7, 2025
Israel’s military surveillance Unit 8200 has reportedly developed a vast database of intercepted Palestinian communications in order to construct an artificial intelligence tool similar to ChatGPT, a joint investigation by The Guardian, +972 Magazine and Mekomit alleged on Thursday. Israel reportedly hopes that the resulting AI tool “will transform its spying capabilities.” The investigation by the ...
- Trojans disguised as AI: cybercriminals exploit DeepSeek’s popularity
March 6, 2025
Among the most significant events in the AI world in early 2025 was the release of DeepSeek-R1 – a powerful reasoning large language model (LLM) with open weights. It’s available both for local use and as a free service. Since DeepSeek was the first service to offer access to a reasoning LLM to a wide ...
- How hackers ruined a Disney employee’s life after he downloaded AI photo tool
February 27, 2025
A former Disney employee’s world was turned upside down when he downloaded an artificial intelligence-powered photo program, unaware that it was laced with hacking software, during a massive data breach at the entertainment giant. In July, Matthew Van Andel, an engineer at Disney at the time, got a message on the chat forum Discord from an ...