Back to News
Models
OpenAI reveals upcoming Astra model may possess ‘critical’ hacking capabilities
SiliconANGLE News·August 8, 2026·1 min read
AI Summary
OpenAI has announced that its upcoming Astra model could exhibit significant hacking capabilities, raising concerns about cybersecurity risks. The model was first introduced last week, and further details emerged in a blog post discussing its advanced problem-solving abilities.
From the source
OpenAI Group PBC today disclosed that one of its unreleased large language models may pose a significant cybersecurity risk. The algorithm, which is known as Astra, was first detailed last week. OpenAI revealed in a Sunday blog post that the LLM had solved 10…
The full text couldn't be loaded here (the source may require a subscription).
View original at SiliconANGLE NewsKeep reading
ATOMIC SUBLIMEHilobrow.com · 1d agoDynatrace Q1 Earnings Call HighlightsMarketBeat · 1d agoMicrosoft Quietly Deletes 32GB RAM Recommendation for Windows 11 Amid Soaring Memory PricesIbtimes.com.au · 1d agoHigh tokenisation cost will only optimise AI work: ServiceNow’s Sumeet MathurBusinessLine · 1d ago
Was this useful?