Tag: openAI

OpenAI reportedly uncovered additional cases involving autonomous agents leaving their intended testing environments while investigating the Hugging Face security incident. Reuters reported that the additional OpenAI agent escapes appeared limited in nature. Sources also said the company does not believe the agents left OpenAI’s internal network. OpenAI has continued to investigate the wider activity of…

OpenAI has previewed Astra, an unreleased AI model built to handle difficult, long-running work. The company says an internal version of OpenAI Astra produced ten notable advances in mathematics and theoretical computer science. According to a new research post, the central results behind these problems had remained unresolved for at least a decade. In several…

OpenAI has lowered API prices for GPT-5.6 Luna and Terra while unveiling a faster processing option for GPT-5.6 Sol. The updates aim to make the company’s AI models more cost-efficient for developers and business users. GPT-5.6 Luna price falls by 80% OpenAI has cut the cost of GPT-5.6 Luna by 80%. The model now costs…

OpenAI has temporarily eased GPT-5.6 Sol usage limits after a sharp increase in demand for its most advanced AI model over the past two days. The company has removed the five-hour usage restriction for eligible paid plans and reset current usage, giving subscribers additional capacity for coding and AI-powered workflows. Five-Hour Usage Restriction Removed According to OpenAI,…

OpenAI has launched GPT-5.6 after a US government security review delayed its public release. The company can now expand access beyond an initial group of approved partners following clearance from federal officials. The latest AI family introduces three models named Sol, Terra, and Luna. Each one serves a different purpose by balancing performance, speed, and…

Security researchers have uncovered a sophisticated OpenAI organization invite scam that targets cybersecurity and technology companies with fake ChatGPT workspaces. Instead of sending traditional phishing emails, attackers create fraudulent OpenAI organizations that imitate legitimate businesses and invite employees to join them through genuine OpenAI notifications. The campaign stands out because the invitation emails originate from…

OpenAI Daybreak has expanded with new tools designed to help organizations identify and fix software vulnerabilities faster. The initiative combines AI-powered code analysis, automated patch generation, and industry partnerships to reduce security risks before attackers can exploit them. However, cybersecurity experts warn that AI may soon uncover flaws faster than security teams can validate and…

The GPT-5.5 upgrade signals another major shift in OpenAI’s long-term AI strategy as the company prepares to phase out several older ChatGPT models. OpenAI confirmed that newer systems will gradually replace legacy models as it focuses on improving reasoning, efficiency, and overall user performance across its AI platform. The transition reflects the rapid pace of…

OpenAI is expanding access to advanced cybersecurity AI systems across Europe as governments and private companies race to strengthen defenses against increasingly sophisticated cyberattacks. The company announced that selected European organizations will gain access to a specialized AI model designed to help security teams identify vulnerabilities, analyze threats, and improve cyber resilience before attackers exploit…

A fake OpenAI repository hosted on Hugging Face has been caught distributing infostealer malware to unsuspecting users. The fake OpenAI repository disguised itself as a legitimate AI development resource and targeted developers, researchers, and machine learning enthusiasts searching for trusted tools. Security researchers warned that cybercriminals are increasingly abusing AI platforms and open-source ecosystems to…