New ‘persona vectors’ from Anthropic let you decode and direct an LLM’s personality
A new study from Anthropic introduces “persona vectors,” a technique for developers to monitor, predict and control unwanted LLM behaviors.Read More
A Single Poisoned Document Could Leak ‘Secret’ Data Via ChatGPT
Security researchers found a weakness in OpenAI’s Connectors, which let you hook up ChatGPT to other services, that allowed them to extract data from a Google Drive without any user interaction.
AI system discovers visual categories while adapting to new contexts
A new approach called open ad-hoc categorization (OAK) helps AI systems dynamically reinterpret the same image differently depending on the categorization context, rather than using fixed visual interpretation. A University of Michigan-led study on this topic was presented in June 2025 at the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) in Nashville, Tennessee.
Anthropic’s new Claude 4.1 dominates coding tests days before GPT-5 arrives
Anthropic’s Claude Opus 4.1 achieves 74.5% on coding benchmarks, leading the AI market, but faces risk as nearly half its $3.1B API revenue depends on just two customers.Read More
US Coast Guard Report on Titan Submersible Implosion Singles Out OceanGate CEO Stockton Rush
A report on the death of five people in the Titan submersible blames design, maintenance, and inspection flaws for its failure. “It all came back to Mr. Rush,” the head of the investigation told WIRED.
OpenAI releases free, downloadable models in competition catch-up
OpenAI on Tuesday released two new artificial intelligence (AI) models that can be downloaded for free and altered by users, to challenge similar offerings by US and Chinese competition.
ChatGPT rockets to 700M weekly users ahead of GPT-5 launch with reasoning superpowers
ChatGPT reaches 700 million weekly users as OpenAI prepares to launch GPT-5 with integrated reasoning capabilities in August 2025.Read More
Science Reveals the Surprising Origins of the Potato
Potatoes as we know them today are the product of a hybridization that took place 9 million years ago between two plants, one of which was an ancestor of the tomato.
Topological approach detects adversarial attacks in multimodal AI systems
New vulnerabilities have emerged with the rapid advancement and adoption of multimodal foundational AI models, significantly expanding the potential for cybersecurity attacks. Researchers at Los Alamos National Laboratory have put forward a novel framework that identifies adversarial threats to foundation models—artificial intelligence approaches that seamlessly integrate and process text and image data. This work empowers …
Read more “Topological approach detects adversarial attacks in multimodal AI systems”
Why tomorrow’s best devs won’t just code — they’ll curate, coordinate and command AI
AI coding requires a serious structural change. Where does that leave entry-level developers and the software industry as a whole?Read More