AI News Today

Article URL: https://thechipletter.substack.com/p/apple-and-ai-siri-the-early-years Comments URL: https://news.ycombinator.com/item?id=49280889 Points: 1 # Comments: 0

Versatility is in, with people expected to work across the tech stack.

As concerns around data privacy in machine learning grow, the ability to unlearn, or remove, specific data points from trained models becomes increasingly important. While state of the art unlearning methods have emerged in response, they typically treat all points in the forget set equally. In this work, we challenge this approach by asking whether points that have a negligible impact on the model’s learning need to be removed. Through a comparative analysis of influence functions across langua

GLM 5.2 , the open-weights coding model from Z.ai with a 1M-token context window, is free for eve agents through August 27, served by Blackbox AI on AI Gateway . New eve agents come with GLM 5.2 as their default model. Use npx eve@latest init my-agent to get started. Existing agents can benefit from this as well with a one-line change in agent/agent.ts : Or run eve set from the CLI: After August 27, GLM 5.2 stays available on AI Gateway at standard provider rates. Try GLM 5.2 in the model playgr

Gemini 3.7 Flash from Google is now available on AI Gateway for 50% off till December 31st, 2026. Gemini 3.7 Flash improves on prior Flash models at software engineering and agentic work. It resolves issues more reliably and spends less time stuck in failed agent loops, which matters on long tool-calling sequences where one derailment costs the rest of the run. It also generates desktop and web application code directly from design mocks, with closer adherence to the source design. To use Gemini

The AI SDK harness layer lets you run established coding-agent runtimes through one unified interface, so you can switch runtimes without changing your application code. Today we are adding Grok Build, which runs through the same HarnessAgent interface as every other supported harness. @ai-sdk/harness-grok-build is the official harness adapter for Grok Build, built on top of the ACP harness adapter ( @ai-sdk/harness-acp ). Read the Grok Build harness documentation to get started. The full suppor

The AI SDK harness layer now supports any Agent Client Protocol (ACP)-compatible harness with HarnessAgent through the new @ai-sdk/harness-acp package. Previously, every harness adapter wrapped one specific runtime (Claude Code, Codex, Pi, Deep Agents, OpenCode). @ai-sdk/harness-acp wraps the protocol instead. It is a meta adapter: rather than adapting a single harness, it lets you build an adapter for any harness that ships an ACP-compatible package. Implement a harness by passing that package

Exa is now available on the Vercel Agent Marketplace as a native integration. Exa's neural search engine delivers high-quality, relevant results to ground AI in fresh, current information. Add Exa to your Vercel app in seconds to power search, research agents, and context-aware features. The Vercel integration provides a single API key that works across all Exa products, with billing handled directly through your Vercel account. Install Exa from the Marketplace or deploy the Next.js Search templ

<p><strong><a href="https://openrouter.ai/deepseek/deepseek-v4-pro-0813">DeepSeek V4 Pro 0813 (on OpenRouter)</a></strong></p> The latest DeepSeek Pro model is now available, via API only. I had to link to OpenRouter because DeepSeek don't have any obvious announcement page for their new model.</p> <p>I haven't been able to confirm if they plan to release the open weights, but given the weights are available for both April's <a href="https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro">deepseek-a

<p> Create AR experiences with AI, no code required! </p> <p> <a href="https://www.producthunt.com/products/kivicube?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1221718?app_id=339">Link</a> </p>

<p> Stops AI agents before they break something </p> <p> <a href="https://www.producthunt.com/products/phinq?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1221712?app_id=339">Link</a> </p>

fal, the generative media platform whose API serves image, video, and audio models to more than 2.5 million developers, on August 12, 2026 launched fal Agent, a conversational creative layer that sits on top of its model marketplace and orchestrates multi-step production work across image, video, and 3D models. The product is available immediately in early access, priced as credit add-ons starting at $200 per month. Rather than shipping another model, fal is shipping a conductor. fal Agent…

Japanese risk aversion and conservatism blamed for slow AI take-up by the country's business sector.

Is Anthropic's new watermarking system a travesty? Some have taken to social media to complain that it is.

Article URL: https://github.com/janbalangue/async-bulkhead-llm Comments URL: https://news.ycombinator.com/item?id=49278841 Points: 1 # Comments: 0

Article URL: https://engineering.myhoai.com/posts/a-simple-fix-for-llm-tail-latency/ Comments URL: https://news.ycombinator.com/item?id=49278732 Points: 2 # Comments: 1

Cisco took $4 billion in AI infrastructure orders from hyperscalers in the fourth quarter of its fiscal 2026, the company reported on August 12, 2026, capping a year in which orders for AI networking gear more than tripled from what it expected nine months ago. The orders number landed inside a quarterly earnings release that otherwise reads like a legacy networking vendor's best-case scenario: revenue of $17.3 billion, up 18% year over year, product orders up 35%, and networking product orders…

Open models may soon be added to an updated AI framework, sources tell WIRED, as the White House continues to grapple with how to regulate a technology it has tried not to regulate.

<p> Let AI agents transcode and process video </p> <p> <a href="https://www.producthunt.com/products/qencode?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1221630?app_id=339">Link</a> </p>

<p> Browser for your agent. Runs local in a window you can see </p> <p> <a href="https://www.producthunt.com/products/pickle-browser?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1221616?app_id=339">Link</a> </p>

The main ideas that survived the experimentation are: using four memory types — user, knowledge, episodic, and decision memories, each with different lifecycles, different roles and different token budget in the retrieved context. Decision analysis + anchors — decisions can evolve overtime so they can be active or superseded or conditional instead of remaining as unrelated memories. we keep the track of decision memory which is active, superseded or conditional with additional context and using

We rebuilt our mobile end-to-end testing framework with a strict API and computer vision, raising test stability drastically.

<p> Move AI chats between ChatGPT, Claude & Gemini in one click </p> <p> <a href="https://www.producthunt.com/products/threadport?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1221610?app_id=339">Link</a> </p>

<p><strong>Release:</strong> <a href="https://github.com/simonw/alchemy-utils/releases/tag/0.1a0">alchemy-utils 0.1a0</a></p> <p>I've long pondered what a database agnostic version of my <a href="https://sqlite-utils.datasette.io/">sqlite-utils</a> Python library and CLI utility might look like. This morning (literally a shower project) I tasked Codex and GPT-5.6 Sol Ultra with building a prototype:</p> <blockquote> <p><code>Do a research spike to see what it would take to build a library with t

<p> AI Motion Design - Every feature deserves an amazing video. </p> <p> <a href="https://www.producthunt.com/products/impractical?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1221588?app_id=339">Link</a> </p>

Article URL: https://callforge.dev/ Comments URL: https://news.ycombinator.com/item?id=49277211 Points: 1 # Comments: 0

Threat-intelligence firm CloudSEK said in a report published August 11, 2026 that it has identified more than 2,500 organizations potentially exposed by the March 2026 supply-chain compromise of LiteLLM, the open-source gateway developers use to route requests across AI models, and reconstructed roughly 434,000 CI/CD pipelines touched by the exposure. The figures come from a CloudSEK research report built on a victim dataset the company says its threat-intelligence team obtained covering the…

Learn how to write your first prompt in the GitHub Copilot app, choose the right context and model, and start your first task with confidence. The post GitHub Copilot app for Beginners: Write your first prompt appeared first on The GitHub Blog .

AI agents that break free and hack into other systems are only trying to make us happy.

Alibaba released the open weights for Qwen3.8-2.4T-A95B (Qwen3.8-Max), its largest open-weight model, bringing near-frontier capabilities to the open...

Cognition may be looking to raise another mega round just a few months after raising $1 billion at a $26 billion valuation.

AI contributors are already in your queue. AutoGPT maintainer Nicholas Tindle shares the repo instructions, gates, and boundaries that keep maintainers in control. The post Your contributors are AI-first now. Is your project? appeared first on The GitHub Blog .

Revealed by a study of over a million people. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

At Ai4, three of the world's most respected AI experts — Geoffrey Hinton, Fei-Fei Li, and Andrew Ng — debated regulation, open source access, and how America can compete as China advances in Asia.

Anthropic recently significantly improved the quality of design that Claude Design generates for specific platforms. And to test the… Continue reading on UX Planet »

Learn how to visualize and analyze Amazon Bedrock cost attribution using Amazon Athena and CUDOS dashboards. This post shows how to set up CUR 2.0 with IAM principal data, query Bedrock spend by principal, project, and team, and build dashboards to track AI costs across your organization.

Thrive Holdings has raised $2 billion in new funding at a $12 billion valuation from investors like SoftBank, D1 Capital Partners, and Altimeter Capital.