OpenAI has released two new open-weight AI models, gpt-oss-120b and gpt-oss-20b, aimed at advancing reasoning in open-source AI. The models are designed for developers, offering powerful capabilities on accessible hardware and are available under the Apache 2.0 license, marking a significant strategic shift for the company.
OpenAI has released two new open-weight AI models, gpt-oss-120b and gpt-oss-20b, aimed at advancing reasoning in open-source AI. The models are designed for developers, offering powerful capabilities on accessible hardware and are available under the Apache 2.0 license, marking a significant...
Chinese robotics firm Unitree has launched the R1, an AI-powered humanoid robot priced under $6,000. The R1 aims to make advanced robotics accessible to educators, researchers, and hobbyists, signaling a major shift toward the consumer market.
A new survey reveals that 42% of developers using AI state that at least half of their codebase is AI-generated. This growing reliance raises concerns about security and quality, as many bypass critical code review processes, highlighting a significant shift in software development practices.
A deep dive into the mind of DeepMind's CEO, exploring his vision for a future shaped by artificial general intelligence – a future of unprecedented progress and profound risks.
OpenAI is reportedly planning to launch its highly anticipated GPT-5 in August, alongside a new "open-weight" model. This dual strategy signals a major shift, combining the development of a premier, closed model with a more open approach to foster competition and innovation in the AI landscape.
New research from Anthropic explores how AI model 'personalities' can shift, influenced by training data. The study introduces 'persona vectors' to monitor and mitigate undesirable traits like 'evil' or sycophancy, aiming to enhance AI safety and control.
Studies reveal that code generated by Large Language Models (LLMs) frequently contains critical security vulnerabilities. Developers relying on AI assistants may produce less secure code while...
A new attack technique, dubbed BYOEDR, allows hackers to exploit free trials of EDR software to disable existing security defenses. Researchers demonstrated how one EDR could be used to neutralize...
New undetectable Plague malware variant targets Linux, evades all antivirus.
The 2025 State of the vCISO Report by Cynomi reveals a more than threefold increase in vCISO service adoption by managed service providers, reaching 67%. Artificial intelligence is a key driver,...
Gartner has released its 2025 Magic Quadrant for Endpoint Protection Platforms, naming leaders like CrowdStrike, SentinelOne, and Microsoft. The report highlights the growing importance of AI,...
A critical "Man-in-the-Prompt" vulnerability threatens popular GenAI/LLM tools like ChatGPT and Gemini. Malicious browser extensions exploit the Document Object Model (DOM) to inject prompts,...
