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:
- Subway reportedly hit by LockBit ransomware – but is it half-baked speculation?
January 23, 2024
Subway has allegedly suffered a data breach at the hands of none other than the notorious LockBit ransomware gang. According to a media report, the ransomware-as-a-service provider added the sandwich makers to its data leak site earlier this week after one of its affiliates made away with gigabytes of sensitive data. Read more… Source: MSN News
- Australia: Russian man Aleksandr Ermakov has been sanctioned over the Medibank data breach
January 23, 2024
The Australian government has used Magnitsky-style sanctions for the first time to punish Russian man Aleksandr Ermakov over what it says is his role in the 2022 Medibank Private data breach. Foreign Minister Penny Wong, Home Affairs Minister Clare O’Neil and Deputy Prime Minister Richard Marles made the announcement on Tuesday morning. But what exactly are ...
- VF Corp’s cyber incident causes data breach of 35.5 million consumers
January 19, 2024
Vans sneaker maker VF Corp said on Thursday the cyber incident that hit the company in December led to a breach of personal data of about 35.5 million consumers, and added that it does not expect a material impact to its financials. The unauthorized activity, detected on Dec. 13, disrupted global customer orders on its e-commerce ...
- Microsoft actions following attack by nation state actor Midnight Blizzard
January 19, 2024
The Microsoft security team detected a nation-state attack on our corporate systems on January 12, 2024, and immediately activated our response process to investigate, disrupt malicious activity, mitigate the attack, and deny the threat actor further access. Microsoft has identified the threat actor as Midnight Blizzard, the Russian state-sponsored actor also known as Nobelium. Beginning in ...
- Carnegie Mellon University hit by cyberattack, informs 7,300 people possibly affected
January 19, 2024
Carnegie Mellon University informed about 7,300 people that their personal information may have been compromised in an August cyberattack that was quietly investigated by law enforcement and the university. The breach impacting one of the nation’s top schools for computing was acknowledged by the university as higher education in general faces a growing assault by digital ...
- Ukrainian hackers steal construction plans for 500 Russian military sites
January 18, 2024
Hackers from the group Blackjack, purportedly affiliated with Ukraine’s SBU security service, have breached a Russian state enterprise involved in construction work for the Russian military, and downloaded over 1.2 TB of data, a Ukrainian law enforcement source told NV on Jan. 18. The data from Russia’s Main Military Construction Directorate for Special Projects included more ...