Latest in Artificial Intelligence

Welcome Gemma 4: Frontier multimodal intelligence on device

Scientists Create Plant That Produces Ayahuasca, Shrooms, and Toad Psychedelics All At Once

Scientists have successfully engineered tobacco plants to produce multiple psychedelic compounds including psilocybin, DMT, and other substances traditionally found in ayahuasca, magic mushrooms, and certain toads. This proof-of-concept breakthrough could reduce harvesting pressure on wild plant and animal species while helping preserve Indigenous cultural practices. The technology demonstrates the potential of synthetic biology to produce complex natural compounds through plant biotechnology.

Holo3: Breaking the Computer Use Frontier

KPMG: Inside the AI agent playbook driving enterprise margin gains

KPMG's latest Global AI Pulse survey reveals a significant disconnect between enterprise AI investment and actual business value generation, with organizations planning to spend an average of $186 million on AI while struggling to demonstrate measurable returns. The report highlights strategies and best practices that high-performing enterprises are using to optimize AI implementations and improve profit margins. Companies need to shift focus from simply increasing AI spending to implementing proven operational frameworks that deliver tangible business outcomes.

Florida University Rolls Out Autonomous Delivery Robots

A Florida university has deployed autonomous delivery robots to operate across its campus. These small-scale robots will handle delivery tasks, potentially improving logistics efficiency and reducing human labor for package transportation on campus grounds.

The Inversion Error: Why Safe AGI Requires an Enactive Floor and State-Space Reversibility

This article examines fundamental structural limitations in current AI systems that cause hallucination and corrigibility problems, arguing that scaling alone cannot bridge the gap between how AI systems operate and what's needed for safe AGI. The author proposes that safe AGI requires an "enactive floor" and state-space reversibility—architectural changes that ground AI systems in verifiable reality rather than relying solely on scaling larger models.

OpenAI Now Valued at $852B After Funding Round

OpenAI has reached a valuation of $852 billion in a recent funding round, cementing its status as one of the most valuable private companies globally. This valuation reflects the continued investor confidence in the company behind the popular ChatGPT platform.

I Tried to Find the ‘Arousal Intelligence’ In An Animated, Augmented Reality Porn Star

A journalist explored a new browser-based augmented reality application designed for adult content, testing its features and capabilities. The article examines how the app integrates AR technology with animated characters to create interactive adult experiences.

Podcast: Inside the AI Slop Propaganda Wars

This podcast episode explores Iran's use of AI-generated content for propaganda purposes, particularly focusing on LEGO-themed disinformation campaigns. The episode also covers various tech and cultural incidents including baseball drama and a notably poorly-designed sex application.

‘BLOCKADE’: The Right Is Using AI Content Scanners to Try to Supercharge Book Banning

Right-wing groups are leveraging AI content scanners like Gemini, ChatGPT, and xAI to identify and challenge books in libraries and schools, potentially accelerating book banning efforts. The use of AI tools allows these groups to scale their content analysis and flag materials more efficiently than traditional review methods. This development represents a new technological approach to book censorship that could significantly impact what books remain accessible in educational institutions.

Microsoft Commits $1B to Thailand's AI future

Microsoft is investing $1 billion in Thailand to develop sovereign AI infrastructure, continuing the trend of major tech companies establishing AI capabilities in specific countries. This commitment reflects the global race to build localized AI systems and strengthen technological presence in key regional markets.

How Can A Model 10,000× Smaller Outsmart ChatGPT?

The article explores how significantly smaller AI models can outperform much larger ones like ChatGPT through extended reasoning and thinking processes rather than raw computational scale. It challenges the assumption that bigger models are always better by demonstrating that efficiency in how a model thinks can compensate for having fewer parameters. The piece suggests that focused problem-solving approaches may be more effective than simply increasing model size.

DeepL’s Borderless Business report reveals 83% of enterprises are still behind on language AI

DeepL's 2026 Language AI report reveals that 83% of enterprises lag behind in adopting language AI solutions, despite widespread AI investments across business functions. The study highlights a gap between overall AI adoption and the integration of AI into translation workflows, indicating that most organizations have not yet modernized their language processing capabilities.

The gig workers who are training humanoid robots at home

Gig workers, including a medical student in Nigeria, are training humanoid robots by recording themselves performing various physical tasks and movements from home. These remote workers use simple equipment like ring lights and head-mounted iPhones to capture video data that teaches AI systems how to replicate human actions and behaviors. This represents a new form of distributed, low-cost labor being leveraged by robotics companies to develop more capable humanoid robots.

What Happens Now That AI is the First Analyst On Your Team?

The article explores how professionals are adapting their careers as AI tools increasingly take on analyst roles traditionally held by humans. It discusses the challenges and opportunities that arise when AI becomes the first line of analysis in data-driven teams, forcing workers to rethink their skill sets and career strategies in a rapidly changing technological landscape.

Hershey applies AI across its supply chain operations

The Hershey Company is implementing artificial intelligence throughout its supply chain operations, moving beyond strategic planning to support real-time, day-to-day decision-making. This trend reflects a broader shift in the food production and logistics industries toward using AI-powered data systems for immediate operational improvements rather than just long-term forecasting.

Falcon Perception

Gradient Labs gives every bank customer an AI account manager

Gradient Labs has developed AI account managers powered by advanced GPT models (GPT-4.1 and GPT-5.4 variants) that automate banking support workflows for customers. The solution leverages mini and nano versions of the language models to deliver low-latency, reliable banking assistance at scale.

Any Custom Frontend with Gradio's Backend

Nvidia Invests $2B In Custom Chip Vendor Marvell Technology

Nvidia has invested $2 billion in Marvell Technology to create a strategic partnership integrating Marvell's XPU chips with Nvidia's NVLink technology. This collaboration aims to enhance custom chip capabilities and interconnect solutions for data center and AI applications.