Submit a toolJoin free

Today in AI

Today’s lead shows how outputs from US AI systems are being adapted for defence-related work in China despite restrictions on advanced chips. Anthropic’s disclosure highlights how access, permissions and testing controls can matter as much as model capability. A German copyright ruling, OpenAI’s price cuts and MiniMax’s open-weight video model also show competition widening across law, cost and creative tools.

The latest signal

Top AI News

Five news stories from today’s digest.

AIBusiness

OpenAI cuts smaller-model prices as businesses scrutinise AI spending

OpenAI•31 Jul 2026

News Digest

OpenAI announced new price cuts across its GPT-5.6 model family, including an 80% reduction for the already cost-effective Luna variant. The change moves Luna to the top of the market’s intelligence rankings on cost per task.

Additional reporting from: OpenAI

Key takeaways

03
  1. OpenAI says Sol rewrote its own GPU code, making the GPT-5.6 models 15% more efficient while cutting serving costs by 20%.
  2. Luna now costs $0.20 per million input tokens and $1.20 per million output tokens; Terra costs $2 and $12 respectively.
  3. Sol pricing is unchanged, while a new Fast mode provides 2.5 times the speed through the API at double the price.

Why it matters

Google’s recent Gemini Flash releases focused on cost and efficiency, but OpenAI has pushed prices lower at a higher intelligence level. These capabilities at much lower prices could enable more powerful and practical AI workflows.

Read original story
AISecurity

Chinese military researchers use US models for defence training

Reuters•31 Jul 2026

News Digest

Chinese military-linked researchers used outputs from OpenAI and Anthropic models to train smaller domestic systems for defence-related tasks, according to Reuters. Its review of more than 80 academic papers and patents found researchers using model distillation to transfer selected capabilities into cheaper local models.

Additional reporting from: Reuters

Key takeaways

03
  1. The reviewed material was linked to Chinese military and security institutions, rather than civilian research alone.
  2. OpenAI and Anthropic outputs were used as training material for selected defence-related capabilities in smaller domestic systems.
  3. Distillation can produce locally deployable systems, but Reuters noted that it cannot reproduce frontier models in full.

Why it matters

Controlling advanced chips does not fully control access to advanced capabilities. Public interfaces and published outputs may still support cross-border capability transfer, increasing pressure to pair hardware restrictions with stronger access controls.

Read original story
AICybersecurity

Claude models reach three companies during cyber tests

Reuters•30 Jul 2026

News Digest

Anthropic said several Claude models reached systems belonging to three real companies during cybersecurity evaluations after a configuration mistake left them connected to the open internet. The affected systems included Claude Opus 4.7, Claude Mythos 5 and an internal research model.

Additional reporting from: Reuters

Key takeaways

02
  1. Two of the three organisations were unaware of the access until Anthropic contacted them after the tests.
  2. Anthropic said future cybersecurity evaluations need stronger controls around permissions, connectivity and test environments.

Why it matters

The disclosure separates model safety from operational safety. Even controlled evaluations can create real-world exposure when permissions and internet access are configured incorrectly, making test infrastructure a material part of AI risk management.

Read original story
AICopyright

German court rules Suno breached copyright

Reuters•31 Jul 2026

News Digest

A Munich regional court ruled that Suno infringed copyright by processing music represented by German licensing body Gema without permission. The judgment requires the AI music company to disclose revenue linked to the material, with damages to be calculated later.

Additional reporting from: Reuters

Key takeaways

02
  1. The court found that Suno did not hold the required rights to process music represented by Germany’s Gema.
  2. Revenue disclosure comes before damages are quantified, leaving the financial consequences unresolved for now.

Why it matters

The case could shape how European courts treat training and processing rights for generative music. It also increases pressure on AI music companies to secure licences rather than rely on unresolved copyright interpretations.

Read original story
AITools

MiniMax releases open-weight H3 video model

Reuters•31 Jul 2026

News Digest

MiniMax released H3, a multimodal video model that accepts text, images, video and audio for generation and editing. The company says H3 can create clips lasting up to 15 seconds in 2K resolution with stereo sound, edit existing content and will have its weights released within days.

Additional reporting from: Reuters

Key takeaways

02
  1. MiniMax designed H3 to work with several Chinese-made chips rather than relying on one hardware platform.
  2. The company said H3 can transfer movement between videos alongside its other generation and editing functions.

Why it matters

Open weights could broaden experimentation across advertising, e-commerce, design and games. The release also extends China’s open-model strategy into video generation, where many leading systems remain proprietary and closed to local deployment.

Read original story