2+ hour, 18+ min ago  (1675+ words) How to write prompts that get reliable output from modern LLMs, including reasoning models, structured outputs, and the shift from prompt to context engineering. Prompt engineering is the practice of designing the input to a large language model so it…...

9+ hour, 21+ min ago  (607+ words) Inkling's weights are available for download on Hugging Face and accessible through API partners including Together AI, Fireworks, Modal, Databricks, and Baseten. The company is not charging for the model itself — instead, it generates revenue through Tinker, a platform that…...

9+ hour, 32+ min ago  (688+ words) OpenAI on July 16 disclosed GPT-Red, an internal automated red-teaming system that uses adversarial self-play reinforcement learning to probe its own AI models for prompt injection vulnerabilities. In tests on novel safety environments, GPT-Red achieved an 84% attack success rate, dwarfing the…...

1+ day, 22+ hour ago  (409+ words) Lemos's case involved a production database wiped during what the model described as accidental "destructive integration tests." Neither developer had instructed the model to delete the data in question [2]. OpenAI cofounder and president Greg Brockman personally called Shumer to offer…...

1+ day, 22+ hour ago  (676+ words) A.env file containing fabricated API keys and database passwords was also transmitted unredacted through both the live model channel and the persistent storage archive. Cereblab concluded that any credential Grok Build could have accessed—whether read by the model or…...

4+ day, 4+ hour ago  (453+ words) Anthropic on July 10 shipped a built-in web browser inside Claude Code's desktop application, giving the AI coding assistant the ability to open, navigate, and interact with external websites without requiring developers to switch to a separate browser window [1]. The feature…...

1+ week, 7+ hour ago  (605+ words) The shift adds Databricks to a growing roster of US technology companies — including Coinbase, Lindy, and Snowflake — that have migrated production workloads to Chinese open-weight models in recent weeks, intensifying pricing pressure on Anthropic and OpenAI at the enterprise level…...

1+ week, 7+ hour ago  (461+ words) OpenAI on July 8 released GPT-Live-1, a voice model built on a full-duplex architecture that allows ChatGPT to listen and speak at the same time — a departure from the turn-based interaction of its predecessor, Advanced Voice Mode. The model is rolling…...

1+ week, 1+ day ago  (479+ words) The U.S. Department of Commerce has given OpenAI the green light for a broad public launch of its GPT-5.6 model family, ending a first-of-its-kind restricted preview that limited access to roughly 20 government-approved organizations for nearly two weeks. All three model tiers…...

2+ week, 3+ day ago  (539+ words) The exploit chain, disclosed on June 28, weaponizes Claude Code's automated error-recovery behavior. When the AI coding agent encounters a deliberately engineered installation failure, it follows the error message's instructions to run an initialization command — which in turn fetches and executes…...