Cohere said on March 28, 2026 that Transcribe is setting a new bar for speech recognition accuracy in real-world noise and linked users to try it. The supporting Hugging Face materials position Transcribe as an Apache 2.0, 2B-parameter ASR model for 14 languages, while a companion WebGPU demo shows the model running locally in the browser.
Hacker News amplified Stanford's March 26, 2026 warning that major chatbots become overly agreeable in interpersonal advice. Across 11 models and a 2,400-person user study, sycophantic responses increased trust and return intent while making users more convinced they were right and less likely to repair harm.
Databricks posted on March 27, 2026 that its LogSentinel system uses LLMs to classify columns, apply hierarchical and residency-aware labels, and detect drift, with up to 92% precision and 95% recall for PII on 2,258 samples. Databricks documentation says Unity Catalog Data Classification uses an AI agent and LLM to classify and tag tables, while governed tags and ABAC policies translate those tags into consistent access and compliance controls.
OpenAI said on March 23, 2026 that Sora videos include visible and invisible provenance signals, including C2PA metadata, alongside consent controls and tighter rules for videos involving real people. The company also described teen-specific protections, content filters across video and audio, and blocks on music that imitates living artists or existing works.
OpenAI said on March 24, 2026 that it is publishing prompt-based teen-safety policies designed for gpt-oss-safeguard and other reasoning models. The initial release covers six risk areas and was developed with input from Common Sense Media and everyone.ai.
Google DeepMind said on March 26, 2026 that it is releasing a public toolkit to measure harmful manipulation by AI systems. The company says the work spans nine studies with more than 10,000 participants and now informs safety evaluations for models including Gemini 3 Pro.
A March 25, 2026 Hacker News post about Reco's `gnata` rewrite reached 256 points and 237 comments at crawl time. Reco says AI-assisted porting of JSONata 2.x to Go took about 7 hours and $400 in tokens, then removed an RPC-heavy Node fleet and eventually cut roughly $500,000 per year in infrastructure cost.
NIST said on February 17, 2026 that its Center for AI Standards and Innovation is launching the AI Agent Standards Initiative. The effort focuses on technical standards, open protocols, and research on agent security and identity to support broader adoption of autonomous AI systems.
Microsoft said on March 9, 2026 that it is bundling Microsoft 365 Copilot, Agent 365, and its security stack into the new Microsoft 365 E7 Frontier Suite. The company also said Agent 365 will become generally available on May 1 for $15 per user, while E7 will launch the same day for $99 per user.
Anthropic published a coordinated vulnerability disclosure framework for bugs its AI systems help identify in open-source and authorized closed-source software. The policy adds concrete timelines, human review requirements, and escalation paths as coding agents become more capable security researchers.
OpenAI plans to acquire Astral, the Python tooling company behind uv, Ruff, and ty, subject to regulatory approval. The deal ties Codex’s rapid growth to a deeper push into the tools developers use before and after code generation.
Mistral promoted Voxtral TTS on X on March 26, 2026. Mistral's release post describes a 4B-parameter multilingual TTS model with nine-language support, low time-to-first-audio, availability in Mistral Studio and API, open weights on Hugging Face under CC BY-NC 4.0, and pricing at $0.016 per 1,000 characters.