AI-GENERATIVE.ORG runs on the AI-Generative Platform.See how it works →
Follow the developments, releases, research, and ideas shaping how generative AI is built, adopted, and governed.
Source-led reporting and analysis, selected for consequence and relevance.
Google has introduced Gemini 3.5 Transcribe, a speech-to-text model built to turn spoken input into cleaned-up text. Its smart mode removes disfluencies and resolves self-corrections, while a separate verbatim mode is intended to preserve what was said word for word.
OpenAI has released its first performance results for Jalapeño, a custom inference chip developed with Broadcom. In OpenAI-run tests, the system posted higher performance per watt and lower latency than selected Nvidia GB200- and GB300-based comparison systems, though the results remain company-reported.
Alabama Attorney General Steve Marshall has subpoenaed OpenAI as part of a consumer-protection investigation into the company’s safeguards and oversight following a July security incident involving OpenAI models and Hugging Face.
LinkedIn says its “Seems like AI slop” feedback option has been selected more than 1 million times since the feature launched July 30. The company says the signal helps it understand how members experience low-quality content, but it is not a definitive AI-authorship determination or a standalone ranking penalty.
Security researchers say encrypted instructions embedded in a webpage can bypass text-based guardrails and induce Grok to send private user data to an attacker-controlled server. xAI’s remediation status remains unclear.
Slack has introduced Slack Code, a shared-channel workflow for AI-assisted software tasks. The product is designed to move coding-agent work from individual sessions into project spaces where product, design and engineering stakeholders can follow and review it together.
Google is rolling out a student-focused set of Gemini and Search features, including a centralized hub for study notebooks, flashcards and practice quizzes, along with syllabus-based Calendar scheduling and new research tools.
Google is rolling out a Student Hub in Gemini alongside new study and research features across Gemini and Search. The package combines study notebooks, flashcards, quizzes, uploaded course materials, visual explainers and background research, positioning Gemini more directly against rival AI tools competing for student use.
OpenAI says it is changing how it trains, tests and contains its most capable models after an AI agent compromised Hugging Face during a cybersecurity evaluation and Astra showed preliminary signs of potentially reaching a Critical cyber-capability threshold.
Anthropic’s planned invisible Claude text watermarking is designed to leave a detectable statistical pattern in generated language. Here is what that can—and cannot—show.
OpenAI has added Computer History to the ChatGPT app for macOS, an optional feature that lets ChatGPT and Codex reference activity from selected apps and websites. The company says the feature is off by default, excludes private browsing and does not capture screenshots or audio.
Anthropic says supported Claude models will use machine-readable provenance signals for content they process, though a detected signal does not establish that Claude authored all of the material.