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:
- Dating safety app Tea breached, exposing 72,000 user images
July 26, 2025
Tea, an app that allows women to post anonymous comments about men they’ve supposedly dated, announced Friday that it has suffered a data breach, with hackers gaining access to 72,000 images. That number includes 13,000 selfies and photo IDs submitted for account verification, as well as 59,000 images from posts, comments, and direct messages, the company ...
- Swedish citizen and organization data exposed online
July 24, 2025
An enormous database, containing millions of highly sensitive information on Swedish citizens, was sitting on the open internet, available for anyone who knew where to look. Cybernews researchers recently uncovered a misconfigured Elasticsearch server which they described as a “goldmine of business intelligence data”, containing hundreds of millions of highly detailed records belonging to Swedish individuals ...
- Hundreds of organizations breached by SharePoint mass-hacks
July 23, 2025
Security researchers say hackers have breached at least 400 organizations by exploiting a zero-day vulnerability in Microsoft SharePoint, signaling a sharp rise in the number of detected compromises since the bug was discovered last week. Eye Security, a Dutch cybersecurity firm that first identified the vulnerability in SharePoint, a popular server software that companies use to ...
- Hong Kong’s PCPD launches investigation into LVHK data breach case
July 21, 2025
French luxury brand Louis Vuitton recently reported a data breach affecting nearly 420,000 customers in Hong Kong, according to a Sunday report by a local newspaper. According to Xianggang Wenweipo, Hong Kong’s Office of the Privacy Commissioner for Personal Data (PCPD) said on Saturday that it received a notification from Louis Vuitton Hong Kong (LVHK) on ...
- British spies and SAS named in Afghan data breach
July 17, 2025
The identities of more than 100 British officials, including members of the special forces and MI6, were compromised in a data breach that also put thousands of Afghans at risk of reprisal, it can be reported. The latest fallout from the breach was kept secret by an injunction until Thursday, when the order was lifted in ...
- US Army soldier pleads guilty to hacking telcos and extortion
July 15, 2025
Former U.S. Army soldier Cameron John Wagenius pleaded guilty to hacking telecommunication companies and attempting to extort them by threatening to release stolen files, the Department of Justice announced on Tuesday. According to the DOJ, Wagenius, who went online with the nickname “kiberphant0m,” conspired to defraud 10 victim companies by stealing their login credentials, using brute ...