Last Week’s AI News #7

Last week in the AI world was full of breakthroughs: Canva opened a new era of creativity, Claude demonstrated signs of autonomous thinking, and Italy passed the first comprehensive AI law in the EU. Google, Anthropic, and Microsoft are strengthening partnerships in infrastructure and innovation, OpenFold3 accelerates drug development, and NVIDIA’s market value surpassed $5 trillion. Each piece of news is a reminder that AI is becoming not just a technology, but an essential business and creative infrastructure shaping the future here and now.

EN

11/3/20256 min read

The week flew by in the blink of an eye, and the world of artificial intelligence is once again buzzing with news. Every day, AI technologies push new boundaries - from creativity and design to biotechnology, education, and even government regulations.

Last week was no exception. Canva announced a completely new era of creation powered by its own AI model. Meanwhile, Claude demonstrated early signs of autonomous thinking, and Italy became the first European Union country to pass a comprehensive AI law. Google, Anthropic, and Microsoft are strengthening partnerships promising faster breakthroughs in infrastructure and innovation.

In short, the AI world is maturing, global players are consolidating, and technologies are increasingly integrating into everyday business and creative life.

Highlights from the week:

  • Canva launched a new era of AI-powered creativity

  • Claude demonstrated early signs of autonomous thinking

  • Google and Anthropic formed a massive TPU partnership to advance AI

  • Italy became the first EU country to pass a comprehensive AI law

  • Google DeepMind established an international mathematics and AI coalition

  • OpenFold3 opens new possibilities for protein structure prediction and drug development

  • Microsoft expands its Discovery platform for regional partners

  • NVIDIA’s market value surpassed $5 trillion

  • And much more from last week in AI

CANVA LAUNCHES A NEW ERA OF AI-CREATED CONTENT

Canva unveiled a suite of new AI solutions - from its own design-focused foundation model to a creative operating system that allows fully editable designs, enhanced video tools, and more.

The Canva Design Model understands design structure and hierarchy, generating fully editable content, and integrates seamlessly with ChatGPT, Claude, and Gemini.

Tools within the creative operating system include Video 2.0 for faster editing, forms, data connectors, email creation, and a 3D content generator.

The Grow module consolidates marketing processes, analyzing successful campaigns, generating brand-consistent versions, publishing to Meta, and tracking results.

Following the 2024 acquisition of Affinity, the program will relaunch as a free, all-in-one creative platform fully integrated with Canva.

Why it matters for businesses:

  • Reduces costs and time by automating design creation and content adaptation across platforms.

  • Ensures brand consistency, enabling teams to generate content aligned with their identity.

  • Combines creative and marketing workflows into a single platform—from concept to publication.

  • Strengthens competitive advantage, allowing businesses to respond to market changes multiple times faster than traditional methods.

CLAUDE SHOWS EARLY SIGNS OF AUTONOMOUS THINKING

Anthropic researchers found that Claude can sometimes distinguish artificially inserted concepts from what it reads, demonstrating limited introspection.

  • Certain concepts were deliberately embedded into Claude, which it detected and recognized as unusual about 20% of the time.

  • When presented with text containing “inserted thoughts,” Claude could accurately replicate content and separate the added concepts.

  • The models also showed the ability to adjust their internal processes when asked to “think about” specific words, indicating a form of self-regulation.

Why it matters for businesses:

  • Improves AI reliability: The ability to detect artificial information reduces errors.

  • Encourages more deliberate generation: Businesses can better control how AI creates text or analyzes data.

  • Opens opportunities for personalization and analytics: AI that understands what it truly “thinks” versus what it receives externally can generate more targeted content and strategies.

GOOGLE AND ANTHROPIC FORM A MASSIVE TPU PARTNERSHIP FOR AI DEVELOPMENT

Google and Anthropic have established a cloud partnership: Anthropic will gain access to up to 1 million Google TPUs in a multi-billion-dollar agreement. By 2026, over 1 GW of AI compute power is planned.

Claude models will run on Google TPU, Amazon Trainium, and Nvidia GPUs, optimizing training, infrastructure, and research.

Claude serves over 300,000 companies, with large clients growing 7× per year, and Claude Code achieved $500M annual revenue within two months. AWS remains a key partner, while Trainium 2 reduces compute costs.

Why it matters for businesses:

  • Massive AI capacity: access to millions of TPUs enables faster, more powerful models.

  • Strategic multi-cloud approach: reduces dependency on a single provider and optimizes costs.

  • Growing client base and revenue: Claude’s rapid adoption shows strong market potential.

ITALY BECOMES FIRST EU COUNTRY TO PASS COMPREHENSIVE AI LAW

Italy is now the first EU country to adopt a full-scale AI law.

  • Up to 5 years imprisonment for harmful AI-generated or modified content (e.g., deepfakes, identity theft).

  • Children under 14 require parental consent to use AI.

  • Several sectors - including healthcare, education, and labor, will face clearer transparency and human oversight rules.

  • National oversight units established: Agency for Digital Italy (AgID) and National Cybersecurity Agency (ACN).

  • Innovation budget: up to €1B in AI, cybersecurity, and related tech support.

Why it matters for businesses:

  • Regulatory advantage and responsibility companies operating in Italy or planning EU operations must meet stricter AI standards.

  • Risk mitigation and reputation, compliant companies avoid sanctions and reputational harm.

  • Innovation opportunities, up to €1B in support for AI and security projects.

  • Operational limits, Italy becomes a testing ground for EU-wide regulations.

GOOGLE DEEPMIND LAUNCHES INTERNATIONAL MATHEMATICS & AI COALITION

Google DeepMind and Google.org, together with Imperial College, IAS, IHES, Simons Institute, and TIFR, launched a coalition tackling complex scientific challenges using powerful AI tools.

Partners gain access to systems like Gemini Deep Think, AlphaEvolve, and AlphaProof, which have already achieved Olympiad-level results and faster algorithms.

Why it matters for businesses:

  • Future innovation, accelerates progress from hypotheses to formal proof chains.

  • Collaboration and efficiency, shared infrastructure lowers experimental costs.

  • Impact across new domains, math and AI synergy drives breakthroughs in physics, biology, and IT.

  • Long-term value, platform fosters continuous innovation year after year.

OPENFOLD3 OPENS NEW POSSIBILITIES FOR PROTEIN STRUCTURE PREDICTION AND DRUG DEVELOPMENT

OpenFold3 predicts protein structures, partners, ligands, and nucleic acids under an Apache 2.0 license. It’s an open platform for drug discovery, enzyme design, and biomedical research, with scalable pipelines via NVIDIA NIM.

Companies can combine the model with their own data, including projects for inverse protein folding, simulations, and ADMET analysis. The platform reduces time and costs from in silico insights to in vitro validation, with standardized benchmarks, community review, and model documentation.

Why it matters for businesses:

  • Faster drug development, reduces time from computational predictions to lab testing.

  • Cost reduction, open infrastructure allows work without massive investments.

  • Competitive advantage, customizing the model with proprietary data creates unique solutions.

MICROSOFT EXPANDS DISCOVERY PLATFORM FOR REGIONAL PARTNERS

Microsoft launched the Discovery platform for regional partners, combining agent teams, knowledge graphs, and HPC to accelerate lab work. Early internal projects from digital material searches to synthetic cooling prototypes within four months, show that agent cycles shorten development time.

The platform connects startups, universities, and infrastructure partners, enabling reproducible local models with shared knowledge. Use cases include plastic recycling catalysts and manufacturing materials. Shared data, experiment templates, and lab journals reduce routine work, while access controls protect sensitive projects.

Why it matters for businesses:

  • Faster product development, HPC and agents accelerate experiments.

  • Cost efficiency, shared infrastructure lowers expenses.

  • Collaboration, startups, universities, and partners can build together.

NVIDIA SURPASSES $5 TRILLION MARKET VALUE

NVIDIA’s market value exceeded $5 trillion for the first time, marking the rise of AI as a central industry. The company increasingly appears as the core of a new industry, with a supply chain optimized for large-scale training.

Optimists see long-term growth; skeptics await proof that capacity translates to sustainable revenue with diverse clients. Business priorities include sustainable orders, improved supply, and maintained margins as infrastructure scales. Monitoring demand for GPUs, networking, memory, and software is crucial. Deployments beyond hyperscalers, disciplined pricing, and service add-ons show business lines maturing at scale.

Why it matters for businesses:

  • AI infrastructure growth, NVIDIA becomes the backbone of the AI industry.

  • Investment signals, sustainable margins and orders indicate stability.

  • Diversification, growing client diversity reduces dependency on hyperscalers.

OTHER HIGHLIGHTS FROM LAST WEEK IN AI

  • OpenAI is considering a 2026 IPO, potentially valuing the company at $1 trillion. CEO Sam Altman calls it “the most likely path given future capital needs.”

  • Perplexity launched Patents, a free AI tool for patent searches, allowing natural language queries across databases and publications.

  • OpenAI released Aardvark in private beta, a GPT-5-powered AI agent that automatically detects, checks, and fixes security flaws in code.

  • Anthropic opened its first Asia-Pacific office in Tokyo and signed a collaboration with Japan’s AI Safety Institute.

  • Figma acquired the AI creative platform Weavy, rebranding it as Figma Weave and expanding capabilities for images, video, animation, and VFX.

  • OpenAI added Character Cameo to the Sora 2 platform, enabling users to create videos with characters, pets, or other figures while maintaining style consistency.

AI is evolving from a technological novelty into critical infrastructure that will soon be indispensable in both business and creativity.

In just a few days, more happens in the AI ecosystem than used to occur in a full quarter, ranging from regulation to fundamental discoveries, from design automation to biotech revolutions. The pace of change is faster than ever.

Each week is a reminder that the future is being built here and now. The question is: are we watching from the sidelines, or are we part of it?

If you want clear, curated AI news without noise, unnecessary summaries, or ads - subscribe to our newsletter.