AIskimIQ
HomeArchiveAbout
← All briefs

Thursday, 2 April 2026

39 articles

Executive Summary

Anthropic's Claude advances autonomous AI agents while facing a second security breach involving leaked source code, while Google reduces costs for its Veo 3.1 Lite video generation tool as regulatory scrutiny of agentic AI systems intensifies globally. Researchers at Caltech also report breakthrough compression techniques for large language models, potentially addressing efficiency concerns in the expanding AI infrastructure landscape.

▶Listen↗
🔬 Research· Model compression

Exclusive | Caltech Researchers Claim Radical Compression of High-Fidelity AI Models

PrismML says its 1-bit large language model achieves radical compression without sacrificing performance, lowering energy consumption.

📰 News· Anthropic security breach

Anthropic leaks its own AI coding tool’s source code in second major security breach

Hundreds of thousands of lines of code were exposed, giving researchers insight into upcoming models and internal architecture.

🔬 Research· Large Action Models

Large Action Models: Hype or Real?

Our research shows that LAM is a large language model (LLM) specifically trained to conduct data-driven actions. See LAMs and LLMs – understand the...

🔬 Research· LLM robotics integration

Combining the robot operating system with LLMs for natural-language control

Over the past few decades, robotics researchers have developed a wide range of increasingly advanced robots that can autonomously complete various...

🚀 Launch· Krafton Raon models

Krafton launched the artificial intelligence (AI) model brand 'Raon' and released four models as ope..

Krafton announced on the 2nd that it has listed four of Raon's first models as open sources on its global platform huggings interface,...

📰 News· governance and security

The agentic shift: When AI becomes the lead hacker

Is our "human-speed" governance already obsolete? This quick dive breaks down the 2025 Anthropic breach, where an AI agent autonomously ran 90% of a major...

🚀 Launch· multi-agent systems

What’s new in Copilot Studio: Updates to multi-agent systems | Microsoft Copilot Blog

Explore what's new in Copilot Studio: Multi-agent systems now generally available, plus updates to the Prompt Editor and governance controls.

🏛 Policy· regulatory guidance

UK regulators warn agentic AI is already here - and it needs watching now

Four UK watchdogs map the risks, rights, and regulatory gaps as autonomous AI agents move from pilot to practice across finance, retail, and communications.

📰 News· computer use capability

Anthropic says Claude can now use your computer to finish tasks for you in AI agent push

Anthropic and its rivals are trying to ramp up capabilities of AI agents after OpenClaw went viral earlier this year.

📰 News· liability and legal risk

Amazon-Perplexity dispute raises questions over AI agent liability

In November 2025, Amazon filed a complaint against Perplexity in the U.S. District Court for the Northern District of California, alleging that use of...

📰 News· agent safety and risks

TCAI Guide: The risks of AI agents built with OpenClaw and other frameworks

Using an AI agent comes with two known risks: The agent has access to all your data and passwords, and it will decide and act on its own.

🏛 Policy· regional framework

Singapore's Agentic AI Framework: Practical Guidance for Market Entry

Singapore continues to position itself as a leading hub for artificial intelligence innovation. In February 2026, Prime Minister Lawrence Wong.

💬 Opinion· large action models

Large Action Models: Hype or Real?

Following the launch of Rabbit, an AI device that can use mobile apps, the term large action models (LAMs) is getting popular. These models move beyond...

📰 News· security and IP

Anthropic Races to Contain Leak of Code Behind Claude AI Agent

Developer issues copyright takedown request in bid to prevent competitors from cloning coding tool's features.

📰 News· defense applications

Startup debuts agentic AI assistant for war

As the Pentagon eyes agentic AI, a veteran-founded company introduces a tool that puts the military first.

📰 News· AI governance and political coalitions

‘Think Everybody Dead’: How the Threat of AI Is Fueling a New Political Alliance

A billionaire-backed movement from Silicon Valley and populists like Bernie Sanders are eager to stop an AI disaster — but can they overcome their mutual...

📰 News· AI safety and security incident

Anthropic Claude Code Leak: AI Giant Exposes Source Code in 'Simple Mistake'; What Was Revealed & Why Experts Are Concerned?

Anthropic Claude Code leak shocks AI industry as source code gets exposed due to packaging error; experts raise concerns over security and AI development...

🔬 Research· interpretability application

Interpretable AI Boosts Cardiovascular Disease Diagnosis

In a groundbreaking advancement poised to redefine cardiovascular health diagnostics, researchers Hasan and Dhrubo have unveiled an innovative artificial...

📰 News· AI safety and security incident

Anthropic has ‘come to copyright’ epiphany after Claude code leak

Source code for Anthropic's AI chatbot, Claude, leaked due to "human error, not a security breach", as the company preached trust and safety.

🚀 Launch· Website creation tool

How to use AI to create a website

From idea to launch, learn how to use AI to create a website and build a website with Copilot using clear steps and prompts.

📰 News· Copilot education expansion

Microsoft invests $5.5B in Singapore AI and Copilot for students | ETIH EdTech News

Microsoft invests $5.5B in Singapore AI infrastructure and expands Copilot access for students and educators. ETIH edtech news explores how AI skills,...

📰 News· AI agent security

Exabeam expands Agent Behavior Analytics to secure AI agents across ChatGPT, Copilot and Gemini

Security intelligence and management solutions company Exabeam Inc. today announced an expansion of its Agent Behavior Analytics platform that adds...

📰 News· AI agent security

Exabeam now monitors AI agents in ChatGPT, Copilot, and Gemini

Exabeam is expanding Agent Behavior Analytics for ChatGPT, Copilot, and Gemini with five new features to combat AI agent threats.

📰 News· Copilot Cowork enterprise

Will MS Copilot Cowork Enable Real Enterprise AI Collaboration?

Copilot Cowork in Frontier has been launched, but the company will need to take specific steps to enable enterprise-grade usage by customers.

📰 News· Copilot research features

Microsoft Copilot Is Using a Surprising AI Trick to Create More Accurate Research Reports

If you're looking for more trustworthy AI-driven research, check out these new Microsoft features.

📰 News· Veo 3.1 Lite and Sora

Google's Veo 3.1 Lite Cuts API Costs in Half as OpenAI's Sora Exits the Market

Google's cheapest video model yet targets developers burned by high generation costs, arriving just days after OpenAI pulled the plug on Sora.

🔬 Research· AI audio generation

The Case Against Text Prompts for AI Sound Generation

AC-Foley shows why text prompts limit video-to-audio generation and how reference audio enables finer control, timbre transfer, and zero-shot synthesis.

📰 News· PixVerse V6

AI Video Generation Tools

AI video generation tools - PixVerse's V6 AI video generation tools introduce a new way to create multi-shot, cinematic content with synchronized audio from...

📰 News· DomoAI update

AI Video and Image Generation Service DomoAI Updated, Easily Create Audio Files from Text

DomoAI, an AI video and image generation platform operated by DomoAI PTE. LTD, has updated its Talking Avatar feature. This allows for the creation of...

📰 News· AI-generated image detection

Viral image of South African shacks made from road signs shows telltale signs of AI generation

In March 2026, the image of an informal settlement showing shacks with walls made from road signs began gaining popularity on Facebook and X in South Africa...

📰 News· industrial AI robotics

NVIDIA and robotics companies are working to deploy AI-powered robots in factories and hospitals

Researchers report that a shared AI software stack is now being tested across industrial robots and surgical systems, marking one of the first coordinated...

🔬 Research· physics-informed ML

Ultra‑robust machine‑learning models run stable molecular simulations at extreme temperatures

Researchers at The University of Manchester have created a physics‑informed machine‑learning model that can run molecular simulations for unprecedented...

🔬 Research· LLM applications in research

AI inspires new research topics in materials science

Researchers at KIT analyze materials science literature; combination of large language models and machine learning indicates trends for future research.

📰 News· market dynamics

VC Hits $297 Billion in One Quarter, AI Swallows 81% of Funding

The first quarter of 2026 marks a historic moment for venture capital. Investors poured $297 billion into 6000 startups worldwide — an increase of around.

📰 News· market dynamics

Q1 2026 Shatters Venture Funding Records As AI Boom Pushes Startup Investment To $300B

The first quarter of 2026 was unlike any other for venture investment, driven by unprecedented spending on AI compute and frontier labs.

📰 News· AMD MLPerf inference performance

AMD Delivers Breakthrough MLPerf Inference 6.0 Results

See how AMD Instinct MI355X delivers breakthrough MLPerf Inference 6.0 results across new GenAI workloads from single GPU to multi-node scale.

🚀 Launch· AI-designed chip design startup funding

Cognichip wants AI to design the chips that power AI, and just raised $60M to try

The firm says it can reduce the cost of chip development by more than 75% and cut the timeline by more than half.

📰 News· Chinese chip market competition

Chinese chipmakers claim nearly half of local market as Nvidia's lead shrinks

Chinese GPU and AI chip ‌makers captured nearly 41% of China's AI accelerator server market last year, eroding Nvidia's once-dominant position in one of its...

📰 News· NVIDIA MLPerf records and hardware optimization

NVIDIA Extreme Co-Design Delivers New MLPerf Inference Records | NVIDIA Technical Blog

Co-designed hardware, software, and models are key to delivering the highest AI factory throughput and lowest token cost. Measuring this goes far beyond...