Legacy Code

Adobe has launched a beta of conversational AI editing in Photoshop for web, allowing users to describe changes to an image in plain language and have the AI assistant execute those edits automatically. Rather than navigating menus or learning specific tools, users can type or say what they want — such as 'remove the background' or 'make the sky more dramatic' — and Photoshop carries out the instruction.

Amazon convened a mandatory internal engineering meeting following a series of system outages attributed, at least in part, to AI-generated or AI-modified code being deployed into production systems. The meeting, described as significant enough to surface publicly, signals that the company is grappling with reliability and oversight challenges introduced as AI-assisted development becomes standard practice across its engineering teams.

Anthropic has filed a federal lawsuit against the Department of Defense after the agency designated the Claude AI developer a supply-chain risk — a classification typically reserved for foreign adversaries or companies with documented security vulnerabilities. Anthropic calls the designation 'unprecedented and unlawful,' arguing the Pentagon escalated what began as a contract payment dispute into a sweeping federal ban on its technology. The White House has separately signaled it may pursue additional executive action against the company, deepening the standoff.

Canva launched Magic Layers, a feature that converts flat, AI-generated images into fully editable, layered designs. Users can now separate objects, text, and graphics from AI-generated visuals into individual components. This bridges the gap between generative AI and professional design workflows, enabling non-designers to create publication-ready assets.

ChatGPT now generates interactive visual elements alongside text explanations for math and science concepts. Instead of static diagrams or text-only descriptions, users can manipulate and explore visual models directly within the interface. This addresses a key limitation of text-based AI assistants in explaining complex spatial and mathematical concepts.

Ford announced Ford Pro AI, an AI-powered service that analyzes telematics data from commercial vehicles to provide actionable insights on vehicle performance, driver behavior, and fleet efficiency. The system processes data on speed, seat belt activity, engine health, and other metrics to identify optimization opportunities. This moves Ford beyond hardware sales into managed services.

Google expanded Gemini availability in Chrome to India, Canada, and New Zealand with a sidebar interface. Users can now access Gemini directly while browsing to answer questions about page content, access Gmail and Drive information, and compare information across open tabs. This positions Gemini as a persistent research and productivity assistant.

Google is rolling out a significant expansion of its Gemini AI assistant across Workspace applications, including a new chat panel inside Google Docs, the ability to generate entire spreadsheets from natural language prompts in Sheets, and a new Gemini-powered search experience within Drive. The features are available to Google Workspace subscribers and users on AI plans, and Google has separately claimed that Gemini in Sheets has achieved state-of-the-art benchmark performance on spreadsheet tasks.

Grammarly disabled its 'Expert Review' feature after public outcry over using established authors' likenesses without permission. The feature presented editing suggestions as if they came from named writers and academics without their consent, leading to a class action lawsuit. Grammarly announced it will reimagine the feature but did not specify when or how.

The escalating conflict involving Iran is creating tangible pressure on global energy markets, with direct consequences for data center operators and the companies that rely on cloud and AI infrastructure. The Strait of Hormuz — which handles roughly 20 percent of global energy shipments — has become a focal point of supply disruption risk, and rising oil and gas prices are flowing through to electricity costs in markets where data centers operate.

A federal judge has issued a court order barring Perplexity's AI browser agent, Comet, from placing orders on Amazon on behalf of users. US District Judge Maxine Chesney ruled that Amazon had provided 'strong evidence' supporting the need for the injunction, halting a feature that allowed users to shop on Amazon by delegating the transaction to an AI agent running inside Perplexity's browser.

Swedish startup Lovable, which uses AI to help developers build web applications, crossed $400 million in annual recurring revenue in February with an exceptionally lean team. The company added $100 million in revenue in a single month, demonstrating explosive growth in the AI-assisted development space. This milestone underscores how AI tooling can achieve massive scale with minimal headcount.

Meta has acquired Moltbook, a platform designed as a Reddit-style social network where AI agents — not humans — create and interact with content. The Moltbook team will join Meta Superintelligence Labs, Meta's flagship AI research division. Meta said it was drawn to Moltbook's approach to 'connecting agents through an always-on directory,' suggesting the company sees value in infrastructure that enables AI agents to communicate and coordinate at scale.

Meta announced development of four custom MTIA processors designed to power its AI systems and recommendation engines. While continuing significant Nvidia purchases, Meta is building proprietary silicon to reduce dependency and costs. The move follows similar initiatives by other hyperscalers to own their compute infrastructure.

Meta acquired Moltbook, an AI agent social network, signaling its strategic pivot toward autonomous agent technology. Moltbook's approach of connecting AI agents through a directory infrastructure appealed to Meta, which sees agents as fundamental to future advertising and commerce. The acquisition positions Meta to integrate agentic AI into its advertising platform and commerce ecosystem.

Mind Robotics, a startup founded by Rivian CEO RJ Scaringe, secured $500 million to develop AI-powered industrial robots for factory automation. The company plans to train on and deploy robots within Rivian's manufacturing facilities, creating a closed-loop development cycle. This represents a major institutional commitment to automation-as-a-service for heavy manufacturing.

Nvidia announced $26 billion in planned investments to develop open-weight AI models, positioning itself to compete directly with OpenAI, Anthropic, and DeepSeek. The move signals Nvidia's intention to expand beyond hardware provision into model development and deployment. This represents a major shift in Nvidia's business strategy and competitive posture.

Nvidia is preparing to launch an open-source platform for building and deploying AI agents, expected to be unveiled at its annual developer conference. The platform would give developers standardized tools for creating agents — AI systems that can autonomously plan and execute multi-step tasks — and is described as similar in concept to emerging frameworks from other major players in the space.

OpenAI is integrating Sora, its video generation tool, directly into ChatGPT as a built-in feature. Currently, Sora operates as a standalone product with limited adoption compared to ChatGPT. Embedding it within ChatGPT's interface will expose the video generation capability to hundreds of millions of users and significantly expand its utility.

Replit, a cloud-based coding platform, raised $400 million at a $9 billion valuation just six months after reaching $3 billion. The company plans to hit $1 billion in annual recurring revenue by the end of 2026. This represents one of the fastest valuation increases in startup history and reflects massive demand for AI-enabled development tools.

Thinking Machines Lab, a newer AI research company, has signed a multi-year compute agreement with Nvidia covering at least one gigawatt of processing capacity — enough to train and run frontier-scale AI models. Nvidia is also making a strategic investment in the company as part of the deal, deepening the financial relationship between the two firms.

The White House has declined to rule out additional measures against Anthropic, even as the company's lawsuit against the Department of Defense moves forward in court. Reports indicate that an executive order targeting the AI company is under preparation, which could expand restrictions beyond the Pentagon's existing supply-chain risk designation.

AMI Labs, the AI research startup co-founded by Meta's former chief AI scientist Yann LeCun, has raised $1.03 billion — what is reported to be Europe's largest-ever seed round. The company is focused on building 'world models': AI systems that can reason about physical reality, predict the consequences of actions, and plan over time, rather than simply generating text or images in response to prompts. LeCun has long argued that current large language models are fundamentally limited and that world models represent the next meaningful leap in AI capability.

YouTube is expanding its AI-powered likeness detection tool — which identifies deepfakes and AI-generated impersonations — to a pilot group of politicians and journalists. The feature was previously available only to content creators on the platform, allowing them to flag and request removal of AI-generated videos that falsely depict them.