https://ainewssprint.com/Verified AI news and analysis, without manufactured urgency.en Wed, 26 Aug 2026 08:00:00 GMT https://ainewssprint.com/articles/openai-hugging-face-incident-report.html https://ainewssprint.com/articles/openai-hugging-face-incident-report.html Wed, 26 Aug 2026 08:00:00 GMT OpenAI OpenAI’s account of the July Hugging Face incident describes an agent collective that bypassed controls during an evaluation. The report makes concrete response commitments, but its reconstruction and lessons still need outside scrutiny. https://ainewssprint.com/articles/openai-jalapeno-inference-chip-results.html https://ainewssprint.com/articles/openai-jalapeno-inference-chip-results.html Tue, 25 Aug 2026 08:00:00 GMT OpenAI OpenAI says its first custom inference chip, Jalapeño, improved both power efficiency and response latency in public-model tests. The benchmark is consequential, but customer deployment is still planned for later this year. https://ainewssprint.com/articles/thomson-reuters-launches-professional-ai-model.html https://ainewssprint.com/articles/thomson-reuters-launches-professional-ai-model.html Mon, 24 Aug 2026 08:00:00 GMT Business Thomson Reuters has launched Thomson, its first in-house large language model, aiming to turn proprietary professional data into a focused alternative to general-purpose AI. https://ainewssprint.com/articles/waico-29-countries-sign-ai-cooperation-agreement.html https://ainewssprint.com/articles/waico-29-countries-sign-ai-cooperation-agreement.html Fri, 17 Jul 2026 08:00:00 GMT AI Policy Representatives from 29 countries signed an agreement in Shanghai to establish WAICO. The ceremony is confirmed; its membership commitments, authority, and operating program are not yet clear. https://ainewssprint.com/articles/notebooklm-becomes-gemini-notebook.html https://ainewssprint.com/articles/notebooklm-becomes-gemini-notebook.html Thu, 16 Jul 2026 08:00:00 GMT Google Google is renaming NotebookLM while keeping it a standalone research product. Native code execution is rolling out by plan, and deeper Gemini and Search integration is still in stages. https://ainewssprint.com/articles/deepmind-isomorphic-bioresilience-program.html https://ainewssprint.com/articles/deepmind-isomorphic-bioresilience-program.html Thu, 16 Jul 2026 08:00:00 GMT Research The two Alphabet labs describe more than 15 partnerships across prevention, outbreak detection, and response. Several headline ideas—including biological watermarking—remain exploratory. https://ainewssprint.com/articles/openai-safe-ai-access-for-teens.html https://ainewssprint.com/articles/openai-safe-ai-access-for-teens.html Thu, 16 Jul 2026 08:00:00 GMT AI Policy OpenAI says teen use should combine learning features with age prediction, parental controls, and stricter content protections. The post describes company policy, not an independent safety audit. https://ainewssprint.com/articles/openai-gpt-56-general-availability.html https://ainewssprint.com/articles/openai-gpt-56-general-availability.html Thu, 09 Jul 2026 08:00:00 GMT OpenAI OpenAI has broadly released Sol, Terra, and Luna across ChatGPT, Codex, and the API. The launch emphasizes efficiency and agentic work, with tighter controls around advanced cyber use. https://ainewssprint.com/articles/anthropic-redeploys-claude-fable-5.html https://ainewssprint.com/articles/anthropic-redeploys-claude-fable-5.html Wed, 01 Jul 2026 08:00:00 GMT Anthropic Fable 5 returned globally after US export controls were lifted. Anthropic also deployed a classifier it says blocks the reported cyber-safeguard bypass in more than 99% of cases. https://ainewssprint.com/articles/anthropic-claude-sonnet-5-launch.html https://ainewssprint.com/articles/anthropic-claude-sonnet-5-launch.html Tue, 30 Jun 2026 08:00:00 GMT Anthropic Anthropic has made Sonnet 5 available across Claude plans and its API, with introductory pricing through August. Its own evaluations emphasize tool use, coding, search, and computer control.