AI news, updated as it matters

The AI news worth your attention.

Short, source-first briefings for the changes shaping how AI is built, used, and governed.

Format

60 words

Coverage

Source-led

Feed

Live

Today in AI

INDAI News Roundup: Watermarks, Drivers, and Model Maps SECOpenAI and NVIDIA Detail AI Security and Infrastructure Safeguards OPELiquid AI Releases LFM2.5 Q4_0 Checkpoints via Quantization-Aware Distillation OPELangGraph SDK 0.4.3 Released with Decrypt Replacement Result POLOpenAI launches democratic oversight initiative, funds policy projects

Live briefing

Top today

See every story

Top todayAug 19

industryAug 191 source

1 sourcePrimaryEvent Aug 19Published Aug 19

OpenAI Reaffirms Zero Data Retention for API Customers

OpenAI reaffirms Zero Data Retention for eligible API customers and previews Private Safety Processing, enabling advanced AI safety without compromising data privacy.

Read the story

Why it matters: Enterprises can use frontier models with stronger data privacy guarantees, potentially easing compliance and security concerns.

industryAug 181 source

ChatGPT Ads Expands to 31 European Markets

OpenAI is expanding ChatGPT Ads to 31 European markets, enabling advertisers to reach users as they explore, compare options, and make decisions within the platform.

1 sourcePrimaryEvent Aug 18Published Aug 18

researchAug 181 source

IBM Research: How Much Memory Does Your Agent Actually Need?

IBM Research introduces ALTK-Evolve-HMM, a method to evolve agent memory architectures using hidden Markov models. The approach automatically designs memory modules, improving performance on long-horizon tasks while reducing memory usage. The work is detailed in a Hugging Face blog post.

1 sourceResearchEvent Aug 18Published Aug 18

industryAug 171 source

Google Gemini and Pixel Partner with Five Football Clubs

Google announced partnerships between Gemini, Pixel, and five global football clubs to enhance fan matchday experiences using AI and smartphone technology. The collaboration aims to bring fans closer to the game through innovative features.

1 sourcePrimaryEvent Aug 17Published Aug 17

open sourceAug 182 sources

Mojo programming language goes open source

Mojo, the Python-inspired language for GPU programming, is now open source. The compiler and toolchain were released under an Apache 2 license, following the 1.0 release last week. Originally planned as a Python superset, Mojo is now its own language, prioritizing GPU programming ease.

2 sourcesIndependentEvent Aug 18Published Aug 18Discussed Aug 18

Browse intentionally

Choose the part of AI you’re following.

Browse all topics

Models

The systems moving fastest

Explore 02

Agents

Useful autonomy, real work

Explore 03

Research

Papers worth understanding

Explore 04

Open source

Tools built in the open

Explore 05

Policy

Rules shaping AI

Explore 06

Chips

Compute and infrastructure

Explore 07

Funding

The business behind the shift

Explore 08

Security

Risks, safeguards, and incidents

Explore

Keep going

Latest across AI.

A continuous, mixed-topic reading flow. Your preferences can shape the order, but never narrow the world you see.

Continue to the feed

open sourceAug 182 sources

Mistral CLI v0.3.0 and Vibe v2.24.2 Released

Mistral released CLI v0.3.0 with OIDC login/logout/whoami commands and CLI-based app evals. Mistral Vibe v2.24.2 adds a session picker, /log-level, /retry for ACP clients, worktree cleanup, faster startup, and multiple bug fixes.

infrastructureAug 171 source

Same Cluster, 33 Points More Utilization: What Changed Was the Order

A Hugging Face blog post describes how reordering tasks on the same GPU cluster boosted utilization by 33 percentage points. The post details scheduling optimizations that improved efficiency without adding hardware, highlighting the impact of job order on cluster performance.

industryAug 181 source

OpenAI Partners with CodeAI to Boost AI Literacy

OpenAI and CodeAI announced a partnership to help students build AI literacy, think critically about AI, and develop skills to use and shape it responsibly. The collaboration aims to prepare the first AI generation for a future where AI is ubiquitous.

researchAug 1234 sources

New Research: LLM Judges Unstable, RAG Privacy, and More

New arXiv papers cover LLM judge instability (Wiggle Framework), privacy-preserving RAG (SEAG), multi-agent pricing, transformer expressiveness, constraint saturation, Dual-Flow Transformers, governed memory, and edge scheduling. Findings: judges flip verdicts 25-91% under pressure; SEAG hides sensitive entities with >74% accuracy; instruction following collapses beyond 5-6 constraints.

industryAug 181 source

OpenAI Launches ChatGPT for Teens with Added Protections

OpenAI introduced ChatGPT for Teens, designed to help teens learn and think critically with AI. It includes stronger built-in protections, healthy-use features, and additional parental controls.

open sourceAug 181 source

OpenAI Codex v0.148.0 Adds Markdown Export, Session Forking, and Amazon Bedrock Support

OpenAI released Codex v0.148.0, adding TUI conversation export to Markdown, session forking with `codex exec fork`, draft prompts during startup, and Amazon Bedrock Runtime as a built-in provider. Bug fixes include model switching, session restoration, and sandbox restrictions failing closed on denied paths.

open sourceAug 181 source

Sentence Transformers Adds Multi-Vector Embedding Models

Hugging Face's Sentence Transformers library now supports multi-vector (late interaction) embedding models, enabling richer document representation. This update allows for more nuanced semantic search and retrieval, improving performance on complex queries.

agentsAug 123 sources

Agentic Coding: From Autocomplete to Autonomous Teammates

Agentic coding tools now autonomously read repos, run tests, and open PRs, shifting from suggestion to delegation. However, a developer warns that vague instructions cause failures, and an orchestration platform analysis shows infrastructure overhead consumes 20-30% of job runtime, with pre-warming caches cutting times from 30 to 20 minutes.

industryAug 181 source

Asana Uses OpenAI Codex to Complete 5 Years of Engineering Work in 2 Weeks

Asana leveraged OpenAI's Codex to replace an outdated testing system in two weeks, a task estimated to take five years, at a cost of about $12,000. The project demonstrates the potential of AI-assisted coding to dramatically accelerate software engineering tasks.

infrastructureAug 31 source

Meta Doubles Training Efficiency of LLM-Scale Ads Model GEM

Meta's Generative Ads Recommendation Model (GEM), the foundation model for ads recommendations on Instagram and Facebook, now trains at LLM scale on thousands of latest-generation GPUs. The company doubled end-to-end training efficiency to 20–25% Model FLOPs Utilization (MFU) while scaling training FLOPs 4x.

industryAug 181 source

NVIDIA Scales Expertise with ChatGPT Work

NVIDIA teams use ChatGPT Work to reduce manual tasks, connect fast-moving signals, and scale successful workflows globally, according to an OpenAI blog post.

open sourceAug 171 source

openai-ruby v0.80.0 adds Amazon Bedrock Runtime support

openai-ruby v0.80.0 released on 2026-08-17 adds Amazon Bedrock Runtime support, plus API updates for Ultrafast tier, structured MCP, websocket errors, and separate websocket events. Bug fixes and chore improvements are also included.

Trending now

Most discussed across AI.

Open trending

OpenAI Reaffirms Zero Data Retention for API Customers

industry · 63 signal

AI News Roundup: Watermarks, Drivers, and Model Maps

industry · 63 signal

OpenAI and NVIDIA Detail AI Security and Infrastructure Safeguards

security · 60 signal

Liquid AI Releases LFM2.5 Q4_0 Checkpoints via Quantization-Aware Distillation

open source · 60 signal

A calmer signal

Source-first

Every card links to the reporting behind it.

Fresh by design

New, eligible stories are prioritised without hiding important context.

Made to scan

The key development, why it matters, and the next source — in moments.