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:
- French naval officer’s jogging app logs Inadvertently expose France’s aircraft carrier location
March 20, 2026
The relentless pursuit of a personal best is a common motivator for athletes, but for one French naval officer, a routine morning run has now been linked to a national security scare. While the French military typically prides itself on stealth and strategic positioning, the precise coordinates of its flagship were recently broadcast worldwide via a ...
- Marquis says over 672,000 people had personal and financial data stolen in ransomware attack
March 18, 2026
Marquis, a technology company used by hundreds of banks to analyze and visualize their customers’ data, says hundreds of thousands of people had their personal and sensitive financial information stolen in a ransomware attack last year. The Plano, Texas-based fintech company is notifying at least 672,075 people that hackers stole their information during the August 2025 ...
- Notorious online data leak market BreachForums taken down by whitehat heroes
March 17, 2026
BreachForums, one of the most popular underground forums for sharing malware, stolen data, and more – was taken down. Now, the admin seems to be giving up and looking for someone to pass the torch to. Over the weekend, the Cyber Counter-Intelligence Threat Investigation Consortium (CCITIC) posted on LinkedIn, saying that both the clearnet and Tor ...
- Swedish government IT system hacked
March 13, 2026
A large amount of sensitive information allegedly coming from a Swedish government IT system has been posted on the darknet, according to Dagens Nyheter and Expressen. DN writes that the newspaper has taken note of the leak and that it appears to contain the source code for a digital identity management system used by several authorities. ...
- Halifax and Lloyds customers hit by online data breach
March 12, 2026
Lloyds, Halifax and Bank of Scotland customers were given access to strangers’ banking transactions in a major online data breach this morning. Customers were able to view charges and payments on their banking apps that were not linked to their own transactions following the suspected technical glitch. Wage payments, HMRC reference numbers and other personal transactions were ...
- Telus probes cybersecurity incident that ‘ShinyHunters’ group claims responsibility for
March 12, 2026
Canadian telecommunications and business services firm Telus is investigating a cybersecurity incident involving unauthorized access to some of its systems, a company spokesperson said on Thursday. The ShinyHunters hacking group told Reuters in a message it stole at least 700 terabytes of data from Telus. All business operations within the company “remain fully operational, and there ...