April 11, 2026 – AI Daily Recap

The AI industry closed out the week with blockbuster infrastructure deals, a cybersecurity model that’s rewriting the vulnerability playbook, and a wave of state-level AI legislation that signals the regulatory tide is rising fast. Here’s everything that mattered.

Meta and CoreWeave Expand AI Cloud Partnership to $21 Billion

CoreWeave and Meta AI infrastructure partnership
Image: CoreWeave

CoreWeave and Meta announced a $21 billion expanded agreement for AI cloud capacity running through December 2032. The deal gives Meta access to NVIDIA’s latest systems, including GB300 and early Vera Rubin platforms, across multiple locations optimized for large-scale AI inference. Combined with a prior $14.2 billion arrangement, CoreWeave now holds roughly $35 billion in total contracts with Meta, making the social media giant one of its largest customers. CoreWeave shares jumped nearly 16% on the news. The deal underscores a broader trend: Meta plans to spend between $115 billion and $135 billion in capital expenditures this year alone — nearly double its 2025 figure — as the race for AI compute capacity shows no signs of slowing.

Anthropic’s Claude Mythos Preview Discovers Thousands of Zero-Day Vulnerabilities

Anthropic Claude Mythos cybersecurity AI model
Image: TechCrunch

Anthropic’s Claude Mythos Preview continues to make waves after its reveal earlier this week. The model, purpose-built for cybersecurity, has already identified thousands of previously unknown zero-day vulnerabilities across critical infrastructure — including bugs in every major operating system and web browser, with the oldest being a 27-year-old flaw in OpenBSD. Through Project Glasswing, Anthropic is giving over 40 companies — including AWS, Apple, Google, Microsoft, and CrowdStrike — access to the model for defensive security work. Anthropic has no plans to make Mythos Preview generally available, acknowledging the model could supercharge cyberattacks if misused.

Perplexity’s Revenue Surges 50% as AI Agent Pivot Pays Off

Perplexity AI agent revenue growth
Image: PYMNTS

Perplexity’s bet on autonomous AI agents is paying off in a big way. The company’s annual recurring revenue jumped 50% in a single month to top $450 million in March, fueled by more than 100 million monthly active users. The growth follows the launch of agent features like Computer and the Comet browser, as Perplexity shifts from AI-powered search to AI that can actually execute tasks. The company now serves tens of thousands of enterprise clients on subscriptions ranging from $20 to $200 per month, and has moved toward usage-based pricing and credit metering for premium tiers.

LM Studio Acquires Locally AI, Bringing On-Device AI to Mobile

LM Studio acquires Locally AI for mobile on-device AI
Image: LM Studio

LM Studio has acquired Locally AI, bringing mobile-native on-device AI capabilities to its popular desktop platform. Adrien Grondin, the creator of the Locally AI apps, is joining LM Studio to lead work on native AI experiences across devices. The acquisition signals a growing push toward seamless cross-device model and agent experiences — run a model on your laptop, pick up where you left off on your phone. For the local-first AI community, this is a significant step toward making on-device inference a first-class experience everywhere, not just on desktop.

Tether Launches QVAC SDK, an Open-Source On-Device AI Framework

Tether QVAC SDK open-source AI framework
Image: SiliconANGLE

Stablecoin giant Tether has entered the AI infrastructure space with QVAC SDK (QuantumVerse Automatic Computer), a fully open-source cross-platform framework for running and fine-tuning AI models locally. Built on a fork of llama.cpp, QVAC integrates text generation, speech-to-text, and on-device translation into a single SDK that works unchanged across iOS, Android, Windows, macOS, and Linux. True to Tether’s blockchain roots, the framework includes built-in peer-to-peer primitives for decentralized model distribution and delegated inference — no centralized infrastructure required.

AI Legislation Accelerates: 25 New Laws Passed Since Mid-March

AI legislation and regulation 2026
Image: Plural Policy

The regulatory picture for AI is shifting fast at the state level. Since mid-March, 25 new AI laws have passed, with another 27 bills having cleared both chambers. Utah Governor Spencer Cox has signed nine AI bills into law this year alone. Meanwhile, therapy chatbot bans are gaining momentum — Maine has sent a ban to the governor, and Missouri is advancing a similar measure through an omnibus health care bill with $10,000 penalties for violations. Alabama’s SB 63, regulating AI use in health care coverage determinations, was approved by the full House on April 8. The pace of legislation suggests that AI governance is moving from debate to action across the country.

The Big Picture

This week’s news crystallizes the two forces shaping AI in 2026: massive capital deployment and mounting pressure for guardrails. Meta’s $21 billion CoreWeave deal and Perplexity’s explosive agent-driven growth show that the industry’s appetite for compute and autonomous AI is only accelerating. But Anthropic’s cautious rollout of Mythos — a model so powerful it found decades-old zero-days yet too dangerous to release publicly — illustrates the tightrope companies are walking between capability and responsibility. Meanwhile, the flood of state-level AI legislation signals that policymakers are done waiting for federal action. And on the open-source side, both LM Studio and Tether are betting that the future of AI isn’t just in the cloud — it’s on every device you own.

Sources

Sign up here!!

Related Articles

AI News Week #15

Week 15 was the week the AI industry stopped warming up. Anthropic crossed $30B in revenue and locked Claude Mythos behind a cybersecurity-only release, Meta dropped $21B more on CoreWeave compute, four frontier-class open-weights models shipped in seven days, 25 new state AI laws passed — and a Molotov cocktail landed at Sam Altman’s front door.

AI News Week #16

Week 16 of 2026 may go down as the most consequential seven days in AI yet: OpenAI shipped GPT-6, Anthropic released Claude Opus 4.7, Stanford’s AI Index declared the U.S. capability lead all but gone, Q1 venture funding hit a record $300 billion, Snap kicked off the AI-driven layoff era — and Meta started building an AI clone of Mark Zuckerberg.

AI News Week #20

Week 20 was the moment the AI era stopped pretending to be a product cycle. Cerebras pulled off the biggest U.S. tech IPO since Uber, Anthropic and OpenAI both repurposed frontier models as cybersecurity weapons, Trump and Xi opened formal AI safety talks in Beijing, and Anthropic quietly overtook OpenAI in enterprise customers.

Responses

Your email address will not be published. Required fields are marked *