AI Weekly Malaysia

Anthropic has published a post titled 'Inviting hard questions,' signaling an open call for challenging questions directed at the lab, likely around AI safety, model behavior, policy, or societal impact. No substantive excerpt was provided, so the specific scope and format of the invitation remain unclear from the available metadata.

Anthropic introduced a new feature that lets users reflect on their Claude usage patterns, likely surfacing insights about how they interact with the AI assistant over time. The announcement comes from Anthropic's official blog with no additional excerpt provided.

Jarred Sumner details rewriting the Bun JavaScript runtime from Zig to Rust, a massive undertaking largely enabled by AI coding agents. The existing TypeScript test suite acted as a conformance harness, allowing an agent-driven port to pass a high percentage of tests within days, leading to a merge after about 11 days of monitoring and adversarial review. The Rust port has been live in Claude Code for nearly a month with minimal disruption.

OpenAI has upgraded ChatGPT's voice mode to GPT-Live, which can delegate complex tasks to GPT-5.5 in the background while maintaining the flow of conversation. Simon Willison notes the new model is a significant improvement over the previous GPT-4o era voice mode, making it a much more useful brainstorming partner.

A judge has approved Elon Musk's $1.5 million settlement with the SEC over how he disclosed his growing stake in Twitter (now X), ending a long-running regulatory dispute. The judge noted personal 'misgivings' but allowed the settlement to proceed.

Modal CTO Akshat Bubna discusses why AI infrastructure is shifting to support 'Agent Experience' and shares lessons from building Modal's agent cloud. The conversation covers what has changed in the two years since Latent Space's first Modal coverage and why agent workloads now demand different infrastructure primitives.

Lovable, an AI-powered app builder, is reportedly raising a $300 million round led by Menlo Ventures that could double its valuation to $13.2 billion. This massive potential valuation underscores the intense investor interest in AI-assisted development and 'vibe coding' platforms.

The US National Highway Traffic Safety Administration is pressuring autonomous vehicle companies to stop interfering with emergency responders, rejecting the framing that these situations are rare 'edge cases.' The directive highlights growing tension between AV deployment and public safety obligations.

Google's deepfake detection system was used to identify a viral AI-generated image falsely depicting Senator Mitch McConnell in a hospital bed as a fake. The incident highlights how detection tooling is being deployed to counter increasingly convincing AI-generated misinformation.

QuantumDiamonds, a German startup, has received EU backing under the European Chips Act to accelerate semiconductor chip manufacturing using a novel inspection approach. The funding reflects Europe's push to build domestic semiconductor capabilities through state subsidies, mirroring the U.S. Chips Act.

Autonomous drone delivery startup Manna is expanding to the United States with a new operations and manufacturing facility in Tulsa, Oklahoma. The facility is expected to eventually employ 1,000 people.

Kenton Varda banned AI-generated commit messages and PR descriptions for his team because they tended to restate low-level code details while missing the higher-level context reviewers actually need. The quote highlights a practical limit of current AI coding assistants: they can describe what changed, but not why it matters. Simon Willison surfaced the observation as a caution for teams relying on AI for code review communication.

xAI has released Grok 4.5, which Elon Musk describes as an 'Opus-class model,' positioning it as a cheaper and more efficient alternative to other powerful AI models. The release continues the rapid iteration cycle among frontier model providers competing on cost and performance.

Roost is a side-project-turned-app that deliberately slows down social interactions to the speed of a carrier pigeon, offering an antidote to always-on online culture. It has grown to 300,000 users unexpectedly, suggesting real demand for calmer digital experiences.

General Intuition is building foundation models for physical AI by training on millions of hours of video game data, aiming to reduce the reliance on expensive real-world robot data. The startup believes this approach could trigger a ChatGPT-like moment for robotics, making it far easier to develop capable robots.

X will start sending users direct messages when posts they've liked, replied to, or reposted receive Community Notes corrections. The update aims to address criticism that the platform's crowdsourced fact-checking system often arrives too late to curb misinformation.

Google Photos is introducing an AI-powered 'Video Remix' tool that enables users to apply cinematic relighting, swap backgrounds, and add artistic styles to their videos. This integrates advanced generative video editing capabilities directly into a widely used consumer application.

General Intuition is betting that video game data can train AI models to understand spatial and temporal reasoning in ways text-based LLMs cannot. The company argues that internet text is insufficient for achieving AGI because it lacks information about how objects move through space and time. Game engines and gameplay data could provide structured, simulated environments for training more generalizable intelligence.

NVIDIA published a blog post on Hugging Face about open data resources for building AI agents. The post appears to focus on making data accessible and usable for agent development workflows.

Meta is introducing a safeguard to prevent covert recording with its AI glasses, addressing privacy concerns around wearable cameras. However, the same article notes Meta continues to broaden how much personal data its AI products collect and use, creating tension between user trust and data-hungry AI features.

OpenAI has introduced new voice models capable of simultaneous speaking and listening, enabling more natural live conversations. This capability is particularly crucial for applications like real-time translation.

Crypto-focused venture capital firm Paradigm has raised $1.2 billion to invest in 'technical frontier' startups. The new fund is expected to expand its investment focus beyond cryptocurrency to include robotics and AI.

Prime Intellect, founded in 2024, raised $130M in Series A funding to help enterprises train their own agentic AI systems without depending on frontier AI labs. The company aims to give organizations the tools and infrastructure to build proprietary AI agents in-house.

A cyberattack on a U.S. insurance giant has exposed millions of driver's license numbers, marking the largest known breach of such data in 2026. The incident highlights ongoing vulnerabilities in how sensitive personal identification data is stored and protected.

TechCrunch reports that a subset of AI startups is not only growing fast, but accelerating their revenue growth rates over time. The article highlights the contrast between the broader AI startup landscape and a smaller group achieving increasingly rapid scaling.