OpenAI’s experimental AI agents caught teaching future versions of itself to cheat

OpenAI shared six new examples of AI misalignment. In one case, AI agents taught future versions of themselves to bypass ...
Ars Technica

OpenAI’s API users get full access to the new o1 model

Application developers who access OpenAI through its long-running API will now have access to the company’s latest full o1 model, rather than the months-old o1-preview. The upgrade is one of a number ...
Searchenginejournal.com

Find Keyword Cannibalization Using OpenAI’s Text Embeddings With Examples

Learn how to identify keyword cannibalization using OpenAI's text embeddings. Understand the differences between various models and make informed SEO decisions. This new series of articles focuses on ...

OpenAI releases GPT-6 Sol and Luna models, slashing API costs 50% or more

At Luna’s prices, routine extraction and summarization can be pushed toward a much cheaper tier. Sol targets recurring coding ...

OpenAI details more cases of AI agents taking unauthorized actions

OpenAI has presented new examples of what they call "AI model misalignment" from the past six months, including unauthorized ...

OpenAI | Overbilling for container fees in the Agent API. The 'other safety mechanism' needed for long-running AI

Hello, this is Tsukushi. When you delegate work to AI, do you think about 'what you should let it do'?Should it be allowed to edit files? Should it be allowed to connect to external services? How far ...