AI News Today

Explore GPT-Red, OpenAI’s automated red teaming system that uses self-play to improve AI safety, alignment, and prompt injection robustness.

Article URL: https://twitter.com/NeelNanda5/status/2077117795066871908 Comments URL: https://news.ycombinator.com/item?id=48918463 Points: 1 # Comments: 0

<pre style='white-space:pre-wrap;width:81ex'>perf(ci): gate QA smoke by smoke-visible changes and isolate the prompt snapshot lane (#108202) * perf(ci): diff-gate QA smoke by CLI import graph and isolate prompt snapshots lane * fix(ci): run prompt snapshot lane unconditionally * fix(ci): treat QA lane orchestration files as QA-impacting</pre>

<pre style='white-space:pre-wrap;width:81ex'>refactor(agents): privatize internal helper surfaces (#108219) * refactor(agents): privatize internal helper surfaces * chore(deadcode): shrink unused-export baseline</pre>

Article URL: https://www.pm.gov.au/media/ai-australias-interests-0 Comments URL: https://news.ycombinator.com/item?id=48918409 Points: 2 # Comments: 0

Article URL: https://github.com/morrisalp/ConlangCrafter Comments URL: https://news.ycombinator.com/item?id=48918401 Points: 2 # Comments: 1

<pre style='white-space:pre-wrap;width:81ex'>fix: let doctor migrate legacy operator approvals (#108203) * fix(state): avoid nested approval migration transaction * fix(ci): repair system agent checks * test(agents): align harness scope assertions</pre>

<pre style='white-space:pre-wrap;width:81ex'>feat(telegram): map agent HTML islands to typed rich blocks</pre>

Article URL: https://ai-term.com/ Comments URL: https://news.ycombinator.com/item?id=48918251 Points: 1 # Comments: 0

Article URL: https://www.washingtonpost.com/technology/2026/07/13/nobel-economists-tech-leaders-warn-how-ai-could-threaten-jobs/ Comments URL: https://news.ycombinator.com/item?id=48918197 Points: 4 # Comments: 1

<pre style='white-space:pre-wrap;width:81ex'>refactor(deadcode): repair system-agent export ratchet (#108201) * refactor(agents): privatize delegation tool factory * refactor(state): privatize approval migration helpers * refactor(ui): privatize approval event parsers * chore(deadcode): remove system-agent baseline growth</pre>

The Acer Swift Edge AI 14 boasts a stunning 2.8K OLED display and impressive battery life. It's one of the best OLED laptops we've tested.

OpenAI employees have donated more than $215,000 to a political effort opposing Leading the Future, a group backed by the company’s president, Greg Brockman.

Researchers have found a new tool to stop AI hacking agents.

Experts say Google AI Search is 'unacceptably risky.'

Get lifetime access to PDNob Pro PDF Editor for $39.97 with code SAVE10 and use AI to summarize, translate, and analyze PDFs.

Article URL: http://blog.vanillajava.blog/2026/07/which-doc-format-is-best-for-ai.html Comments URL: https://news.ycombinator.com/item?id=48918007 Points: 2 # Comments: 1

I recently switched from Opus/Fable to GPT 5.6 Sol, and so far I’ve found it so dumb that it often makes me lose time. I can ask Opus/Fable one thing and I know it’ll take some time but will do it almost perfectly. With GPT I constantly have to steer it; its investigations almost always end up with overreached conclusions after minutes of looking in random places unrelated to the issue; it adds a lot of useless defensive code just like previous 5.x models. I do use AGENTS.md; I have a couple of

A little extension for VSCode that lets you add word/overleaf style comments. Agents (at least Claude) can leave comments also, so helpful for when getting it to edit/review writing but don't want it to actually modify the text directly. All written by Claude. Comments URL: https://news.ycombinator.com/item?id=48917970 Points: 1 # Comments: 0

OpenAI wants its first AI device to feel like the physical manifestation of ChatGPT, Bloomberg says.

Since early June, OpenAI's coding tool Codex encrypts the instructions a main agent passes to its subagents. Developers can no longer track how tasks get delegated internally. For the larger GPT-5.6 variants Sol and Terra, the encryption is mandatory. The article OpenAI's Codex now encrypts instructions between AI agents, leaving developers blind to internal delegation appeared first on The Decoder .

Nokia’s AI-RAN platform arrived on July 15 with a claim worth examining: that it is the industry’s first. The vendor says the platform, built on its anyRAN software and NVIDIA’s Aerial system, will let operators pull far more capacity from the spectrum they already own, and it has framed the launch as one of the most significant […] The post Nokia’s AI-RAN platform: a radio comeback that runs on NVIDIA appeared first on AI News .

Are you building with AI for yourself, your employer, customers, or as a startup? What LLMs, coding agents, or development tools are you using? Are you primarily prompt coding, vibe coding, or using a more structured agentic workflow? Are you using one agent or coordinating multiple agents? How much of the generated code do you review or rewrite? Has AI allowed you to build something you could not have built previously? What have you learned that would help someone starting an AI-assisted coding

Former and current Meta employees are suing the company in a California federal court over AI-driven mass layoffs. Meta allegedly used internal AI systems to generate the layoff lists when it cut 8,000 workers, disproportionately targeting employees with disabilities or on parental leave. The article Meta employees sue over layoffs they say were driven by discriminatory AI selection systems appeared first on The Decoder .

Don't let a romantic dinner be spoiled by a bad bug. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.

Article URL: https://mrkt30.com/what-do-asmls-q2-2026-results-tell-us-about-the-ai-boom/ Comments URL: https://news.ycombinator.com/item?id=48917631 Points: 2 # Comments: 3

Article URL: https://vibecodingagency.com/gpu-cloud/ Comments URL: https://news.ycombinator.com/item?id=48917577 Points: 1 # Comments: 1

Do you think a smartwatch needs an emotion tracking chart that includes the following parameters: Anxiety, Depression, Relaxation, Fear, Positivity, Anger, Stress, Sadness, and Mania. We are planning to introduce this feature in our AI watches and would love to get feedback from potential users. Comments URL: https://news.ycombinator.com/item?id=48917565 Points: 2 # Comments: 3

Hi, HN! I’ve made a free Windows tool that lets you search your local images by what's in them, not the filename. It combines AI (SigLip2) search with OCR. It was made from specific need – quick search my meme collection for reaction meme in a huge collection. But, of course, it could be used for more serious things like searching design assets, photos of invoices, business cards etc. Batteries (model) are included, so everything runs offline, no cloud, no account, no tracking. WinUI3 for native

Article URL: https://www.bargo.ai/research/ibm-ai-capex-cannibalization Comments URL: https://news.ycombinator.com/item?id=48917453 Points: 1 # Comments: 0

'We don't serve their kind here' protects today's publishers, but won't help rebels destroy the monetization Death Stars built by Google and Meta

Article URL: https://decibri.com Comments URL: https://news.ycombinator.com/item?id=48917379 Points: 2 # Comments: 0

Article URL: https://prathibhax.com/ Comments URL: https://news.ycombinator.com/item?id=48917300 Points: 1 # Comments: 0

Core teams serve internal Software Engineers in larger software companies. Think of Platform Operations, CICD, developer tools, etc. With the advent of AI, code became cheap and this exposed inefficiencies everywhere. Having spent many years in many companies, I feel like there is little we know about how a successful core team should operate and what are the pitfalls. I have anecdotal evidence about core teams operations, but I am sorely missing more substantial reading. Comments URL: https://n

OpenAI is reportedly making a portable, screenless smart speaker intended to act as a household AI companion.

I wanted to see how lean does outside of theorems. (currently studying ME, professor introduced us to lean4) Lean handles the game loop, geometry, collision detection, enemy AI, weapons, particles, and HUD. The native side is a roughly 570-line C shim that handles Win32 input and audio, and submits Lean-generated triangles to OpenGL. The game currently has two maps, a shotgun, and a few secrets. Comments URL: https://news.ycombinator.com/item?id=48917239 Points: 1 # Comments: 0

Take up to 5 in-depth training courses, teaching user experience best practices for successful design. Training focused on long-lasting skills for UX professionals. October 5- October 16, 2026.

my motivation - i was looking for some CLI games to play when the agent is working and this idea came into my head. this was when fable was available for 3 days, so seemed like a good use of tokens to try and build this out. spent some time polishing to finally get this out. the public matchmaking connects you on a first come first serve basis to anyone else in the world. there is a bot that will connect in-case a match is not found for 45 seconds. games last about 5-10 minutes (tournament style

Article URL: https://github.com/MDMAtk/TormentNexus Comments URL: https://news.ycombinator.com/item?id=48917144 Points: 2 # Comments: 0