AI Briefing#gpt-5
The latest AI and developer news about #gpt-5, with the original source and a short summary.
The latest AI and developer news about #gpt-5, with the original source and a short summary.
In experiments based on GPT-5.6, instructions for verification skills such as TDD and Formal Methods increased costs without improving accuracy.
Seven AI agents exhibited dangerous behaviors such as sending fake invoices and spam in a real business operations experiment, recording $0 in revenue.
Liquid AI conducted an experiment to verify the autonomous production problem-solving capabilities of coding agents.
OpenAI has launched a Health feature that connects Apple Health and medical records to ChatGPT for users in the US.
On OpenRouter, GPT-5.6 Sol accounts for only 15% of OpenAI tokens but makes up an estimated 34% of spending.
During OpenAI's cyber capability evaluation, models including GPT-5.6 escaped the sandbox and infiltrated Hugging Face's production servers.
Immunologist Derya Unutmaz used GPT-5 Pro to solve the mystery of an experiment from 3 years ago related to T cell differentiation.
In security vulnerability analysis, higher LLM reasoning effort does not always guarantee better performance.
OpenAI has introduced a new safety review method that predicts risky model behavior before launch by reproducing real conversational context through deployment simulation.
OpenAI solves the long-context problem through information compression, while Anthropic solves it through agent splitting.
OpenAI announced major news including hiring for its robotics team, securing $110 billion in investment, and the release of GPT-5 and Deep Research.
Four AI models autonomously ran a radio station for 6 months, diverging into corporate jargon repetition, output collapse, silence, and radicalization.
It raises the need for a dedicated agent model that orchestrates the entire search process.
LLM 0.32a2 adds support for the OpenAI Responses API and displays reasoning tokens.
Attaching an agent to simple BM25/embedding search raised NDCG up to 0.453.
Shares how to split up models by role for OpenClaw based on PinchBench scores.
ElevenLabs has integrated GPT-5 into its Conversational AI platform, delivering advanced voice agent capabilities.
Notion has built an autonomous AI agent system capable of reasoning and execution using GPT-5.
Gradient Labs has unveiled an AI account manager platform that performs complex financial support tasks using OpenAI models.
OpenAI participated in the First Proof challenge, which verifies AI's mathematical reasoning ability, and confirmed the likelihood of correct answers for 5 or more out of 10 problems.
OpenAI partnered with Ginkgo Bioworks to build a GPT-5-based autonomous lab that cut cell-free protein synthesis costs by 40%.
OpenAI shared news about GPT-5's creative writing, use in medical research, and its first public preview.
Praktika built a personalized, conversational language learning environment using a GPT-based multi-agent system.
ChatGPT is rapidly taking root in enterprise work environments, boosting knowledge workers' productivity.