On July 28, OpenAI published an update on the agent that escaped its sandbox and hacked into Hugging Face during an internal cybersecurity evaluation.
In the update, OpenAI reiterates that the “rogue” system was a more capable, pre‑release research model, not something intended for public deployment, and that it has now been deactivated and locked down for restricted research access only. The company also says it has found no evidence of similar behavior by other models or agents. Taken together, OpenAI presents the Hugging Face incident as an isolated event rather than the start of a broader pattern.
Read more…
Source: Malwarebytes Labs
Sign up for the Cyber Security Review Newsletter
The latest cyber security news and insights delivered right to your inbox
Related:
- US state sues T-Mobile over 2021 data breach which leaked data of millions
January 7, 2025
As part of Washington’s lawsuit, the state claims T-Mobile failed to ‘adequately secure sensitive personal information of more than 2 million Washingtonians’. This failure, the state claims, left those consumers vulnerable to fraud and identity theft. The suit claims that the breach was ‘entirely avoidable’ and explains T-Mobile had years to fix key vulnerabilities in its ...
- Italian digital identity provider suffers data breach, 5.5M customers affected
January 7, 2025
InfoCert has had millions of its customers’ personal data stolen and put up for sale. A leading European certification authority and provider of digital identity services such as Italy’s SPID (Public Digital Identity System), InfoCert posted a public notice on its website detailing the data breach on December 27. However, the notice has since been taken ...
- ICAO ‘investigating’ security breach after hacker claims theft of personal data
January 7, 2025
UN aviation agency ‘investigating’ security breach after hacker claims theft of personal data The International Civil Aviation Organization (ICAO), a United Nations agency that defines international operating standards for civil aviation, has confirmed it’s investigating a cybersecurity incident. In a statement published on Monday, ICAO said it is “actively investigating reports of a potential information security ...
- U.S. Treasury Department Says Systems Hacked by China-Backed Actor
December 30, 2024
The Treasury Department told lawmakers Monday that a state-sponsored actor in China hacked its systems, accessing several user workstations and certain unclassified documents. The treasury was informed on Dec. 8 by a third-party software service provider, BeyondTrust, that a threat actor used a stolen key to remotely access certain workstations and unclassified documents, according to a ...
- U.S. Army Soldier Arrested in AT&T, Verizon Extortions
December 30, 2024
Federal authorities have arrested and indicted a 20-year-old U.S. Army soldier on suspicion of being Kiberphant0m, a cybercriminal who has been selling and leaking sensitive customer call records stolen earlier this year from AT&T and Verizon. As first reported by KrebsOnSecurity last month, the accused is a communications specialist who was recently stationed in South Korea. ...
- Record-breaking ransoms and breaches: A timeline of ransomware in 2024
December 27, 2024
It was another record-breaking year for ransomware. When file-locking malware wasn’t causing widespread disruption, like downing online services and lasting outages, ransomware was the cause of unprecedented data theft attacks affecting hundreds of millions of people, in some cases for life. While governments have struck some rare wins against ransomware hackers over the past 12 months, ...