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:
- Medibank hackers reportedly release all data on dark web
December 1, 2022
Hackers who breached Medibank’s systems have dumped another batch of data on the dark web, along with claims the files contain all of the data they took in a heist that impacted 9.7 million customers. The Australian insurance group confirms six zipped files of data have been released, while government officials reiterate the overdue need ...
- Lastpass says hackers accessed customer data in new breach
November 30, 2022
LastPass says unknown attackers breached its cloud storage using information stolen during a previous security incident from August 2022. The company added that, once in, the threat actors also managed to access customer data stored in the compromised storage service. “We recently detected unusual activity within a third-party cloud storage service, which is currently shared by both ...
- Major Twitter hack sees 5.4 million phone numbers and email addresses leaked on the dark web
November 28, 2022
More than 5.4 million Twitter user records, including personal phone numbers and email addresses, are up for grabs on the dark web in a massive data dump that some believe the Elon Musk-owned firm is attempting to cover up. The data dump was identified by Chad Loder, the founder of cyber security awareness company Habitu8, who ...
- Meta fined €265m over data protection breach that hit more than 500m users
November 28, 2022
Facebook’s owner has been fined €265m (£230m) by the Irish data watchdog after a breach that resulted in the details of more than 500 million users being published online. The Data Protection Commission (DPC) said Meta had infringed two articles of the EU’s data protection laws after details of Facebook users from around the world were ...
- Ransomware gang targets Belgian municipality, hits police instead
November 26, 2022
The Ragnar Locker ransomware gang has published stolen data from what they thought was the municipality of Zwijndrecht, but turned out to be stolen from Zwijndrecht police, a local police unit in Antwerp, Belgium. The leaked data reportedly exposed thousands of car number plates, fines, crime report files, personnel details, investigation reports, and more. This type of ...
- Australian Cyber Task Force Looks to “Hack the Hackers” After Data Breach Crime Wave
November 24, 2022
A recent string of data breaches has prompted rapid changes to Australia’s cybersecurity and data protection policies, and the latest development appears to be a cyber task force set to “hack back” and actively pursue what Minister for Home Affairs Clare O’Neil described as “scumbags.” Home Affairs is promising a new “tough on crime” policy toward ...