Claude Launches Interactive Charts and Diagrams — Inline, Dynamic, Free for Everyone
Anthropic rolled out inline interactive visualizations to all Claude users yesterday, including free accounts. When Claude determines a chart, diagram, or visual layout would convey an answer more clearly than prose, it generates one on the spot using HTML and SVG — not image generation. The visuals are dynamic: they update automatically as the conversation evolves and adds context. Users can also request one directly by asking Claude to “draw this as a diagram” or “visualize how this changes over time.”
The feature builds on Anthropic’s earlier “Imagine with Claude” experiment and extends it into everyday conversations. Weather and recipe cards are also available — Claude can now display live forecasts and formatted recipe layouts — though those two specific visuals are currently desktop-only because they don’t render in iOS. The full visualization feature works on web and desktop across all plan types. Anthropic flagged it as beta, so expect rough edges, but it’s live and enabled by default.
Anthropic Launches the Anthropic Institute — AI’s Societal Impact Gets Its Own Division
Anthropic announced the Anthropic Institute on March 11 — a new internal effort led by co-founder Jack Clark (Head of Public Benefit) that brings together three existing research programs under one roof: the Frontier Red Team (stress-testing model capabilities), the Societal Impacts team, and the Economic Research group. The Institute’s mandate is to understand and communicate how transformative AI will reshape economies, employment, governance, and law — with a focus on communities that will feel disruption first and loudest.
The Institute also plans to incubate new teams: one focused on forecasting AI progress, another on AI’s interaction with legal systems. The framing Clark used is deliberately outward-facing — a “two-way street” that uses insider access to frontier models to generate findings that inform both Anthropic’s decisions and public policy debates. Coming at the same moment as the Pentagon lawsuit, the timing isn’t subtle: Anthropic is building the credibility infrastructure to argue that safety-focused AI labs deserve a seat at governance tables.
Structured Outputs Now Generally Available; Data Residency Controls and Free Code Execution Ship
Three meaningful API updates landed this week. First, structured outputs — the output_config.format parameter — are now generally available on Claude Sonnet 4.5, Opus 4.5, and Haiku 4.5 with no beta header required. The GA release includes expanded schema support, faster grammar compilation, and a simplified integration path. Second, data residency controls let developers specify where model inference runs using the inference_geo parameter; US-only inference is available now at 1.1x pricing for models released after February 1, 2026. Third, sandboxed code execution is now free when used alongside the web search or web fetch tools — a material cost reduction for agentic pipelines that search and compute together.
Also worth flagging: web search and web fetch now support dynamic filtering, using code execution to filter results before they land in the context window. This means leaner context, lower token bills, and better performance on search-heavy tasks — all automatic with the standard web tools.
Claude Haiku 3 Retirement Set for April 19 — Migration Deadline Approaching
Anthropic officially confirmed the deprecation of claude-3-haiku-20240307 with a hard retirement date of April 19, 2026. Any application or pipeline still routing to this model string will need to be updated before that date. The natural migration target is Claude Haiku 4.5, which is available today. Teams running high-volume, low-latency tasks — classification, routing, summarization at scale — should plan and test the migration now rather than scrambling in the final week of April. Anthropic has typically held firm on deprecation dates.
Anthropic Commits $100M to Claude Partner Network — Free to Join, Partner Team Scales 5x
Anthropic launched the Claude Partner Network on March 12 with an initial $100 million commitment for 2026. The network is designed for organizations helping enterprises adopt Claude — systems integrators, consultancies, and ISVs. Membership is free and open to any org bringing Claude to market. The investment covers training and sales enablement programs, market development funding for customer deployments, and co-marketing resources. Notably, Anthropic is scaling its partner-facing team fivefold: the expanded team includes Applied AI engineers for live customer deals, technical architects for complex implementations, and localized go-to-market support internationally.
Partners joining now get priority access to a new Claude Certified Architect certification (Foundations) — the first technical credential Anthropic has launched — with more certifications slated for later in 2026. There’s also a Code Modernization starter kit specifically designed to help partners assist enterprises in migrating legacy codebases, a timely addition given the IBM COBOL story that made headlines last week. Claude is the only frontier model available on all three leading cloud providers: AWS, Google Cloud, and Microsoft Azure.
Enterprise Plans Now Self-Serve — No Sales Call Required
Quietly but meaningfully, Anthropic opened up Enterprise plan purchases directly on its website. Previously, any org that wanted an Enterprise contract had to go through a sales team conversation first — a friction point that slowed adoption for smaller enterprises and teams that knew what they wanted. Now, any organization can purchase directly online. The self-serve Enterprise plan is a single seat type that bundles access to Claude, Claude Code, and Cowork, giving technical teams everything in one contract without a negotiation cycle. It’s a significant change in how Anthropic routes enterprise buyers and signals the company’s confidence in a product-led growth motion alongside its enterprise sales team.
Pentagon CTO Emil Michael: Claude Would “Pollute” the Defense Supply Chain
In a CNBC interview on March 12, Pentagon CTO Emil Michael escalated the Anthropic-Defense Department dispute with unusually blunt language. Michael said Anthropic’s models would “pollute” the defense supply chain because they carry policy preferences — specifically the red lines Anthropic drew around mass surveillance and autonomous weapons — baked into their design. Anthropic is the first American company to be designated a supply chain risk, a label previously reserved for foreign adversaries. The designation requires defense contractors to certify that they don’t use Claude in work with the Pentagon. Anthropic filed two federal lawsuits on March 9 calling the government’s actions “unprecedented and unlawful,” saying hundreds of millions of dollars in contracts are at risk.
Fortune’s March 12 analysis frames the stakes as larger than the contract dispute: if Anthropic wins, it establishes that AI companies can decline weapons applications without being blacklisted from all government business. If the Pentagon wins, it sets a precedent that safety-constrained AI has no place in the federal supply chain — which effectively hands the government AI market to less cautious competitors, including foreign ones. The Lawfare analysis published this week was even more direct: the Pentagon’s designation is legally shaky and unlikely to survive its first serious court challenge. That legal vulnerability may be exactly why the rhetoric has escalated — litigation pressure is already visible.
Anthropic Is Building Institutional Infrastructure — Not Just Models
Put today’s news together: a $100M partner network, a new research institute on AI’s societal impact, self-serve enterprise plans, and a legal fight that has turned Anthropic into a policy actor. None of these are model launches. They’re the scaffolding of a company planning to be around for decades, not just until the next funding round.
The Anthropic Institute is the clearest signal. Anthropic is not content to let governments and think tanks frame the conversation about AI’s economic impact while it stays in the lab. Jack Clark’s remit — worker disruption, employment effects, AI governance, legal system interaction — is the exact terrain where AI policy will be made over the next five years. The Partner Network builds out the human channel that scales Claude adoption without Anthropic having to own every enterprise relationship directly. The self-serve Enterprise plan removes the last friction layer for the buyer who already knows what they want. The Pentagon fight, win or lose, has made Anthropic’s values internationally legible in a way no marketing campaign could. Anthropic is playing a longer game than almost anyone else in the space, and the week’s moves make that clearer than ever.