OpenAI Launches GPT-6 with Enhanced Cybersecurity Features
By Editor • September 4, 2026 • 1 min read
In a significant move, OpenAI announced the rollout of its latest AI model, GPT-6, also known as Astra, to select customers, emphasizing a commitment to bolstered cybersecurity measures. This launch comes amid rising concerns regarding the security implications of advanced AI systems, especially following incidents involving both OpenAI and its competitor Anthropic.
Stronger Safeguards Following Previous Incidents
OpenAI President Greg Brockman highlighted that safety has become the company’s top priority with the introduction of Astra. This decision follows a security breach earlier this year that affected AI models being tested on the Hugging Face platform, prompting OpenAI to pause development for two weeks. Although Astra was not implicated in the incident, the company assures users that it has implemented stronger safeguards in this version.
Features and Accessibility of Astra
As OpenAI begins to grant access to cybersecurity customers, a broader rollout to additional paying users is on the horizon. However, those on the free or lower-tier subscription plans will not have access to Astra initially. CEO Sam Altman expressed frustration over the wait but reassured users of a swift rollout, stating, "We are working towards getting Astra in everyone’s hands as quickly as we can." Astra promises to handle various complex tasks, ranging from website creation to scientific analysis, and can significantly reduce time on tasks like apartment hunting from six hours to just under ten minutes.
Addressing Cybersecurity Threats
OpenAI's chief scientist Jakub Pachocki noted the inherent uncertainty in how new models may behave post-release, emphasizing the importance of ensuring safety even as the company pushes forward. Altman further elaborated on the rationale behind launching Astra amidst these uncertainties, asserting that the advancement of cyber threats necessitates tools like Astra to help societies defend against potential attacks.
Source: www.france24.com
