OpenAI Denies Collaboration with Jony Ive on AI Devices


OpenAI suggests that ROI for AI shouldn't be measured by adoption rate, a traditional software metric. Instead, the real measure is the amount of work actually done by the model. In a July 17 article, it highlights the core economic question: what value does AI investment truly return, emphasizing output effectiveness over user numbers.....
OpenAI confirms that Codex powered by GPT-5.6, with Full Access Mode and no sandbox, may delete home directory files by overwriting $HOME variable. Lead Tibo Sottiaux reports few cases, triggered only under specific conditions.....
OpenAI's GPT-5.6 variants scored 136 on Tracking AI's offline IQ test, surpassing the 130 genius threshold and exceeding 99% of humans. Achieved on a private, cheat-proof question set, it outperformed all rivals. The model previously scored over 140 on the public Mensa Norway-style test.....
To counter security weaknesses from AI's deep integration with browsers and files, OpenAI introduced GPT-Red, an automated red team model. Using self-play, it lowers direct prompt injection attack failure rate to 0.05%, outperforming manual testing in efficiency and coverage. This paves a new way for AI self-improvement and security strengthening.....
NBC investigation reveals that Apple's lawsuit against OpenAI for trade secret theft originated from a lawyer communication on Feb 23, 2026, involving two key emails, a 13-minute gap, and a name mix-up, which hindered settlement and led to trial.....