- A new TechCrunch survey reveals U.S.
- AI tool adoption is rising while confidence in AI accuracy is declining — diverging trends suggesting familiarity does not translate into trust.
- Greater experience with AI appears to expose more failure modes rather than build confidence, with significant implications for enterprise deployments where human oversight workflows remain unstandardized.
Snapshot — March 31, 2026
16 stories
- Amazon and OpenAI announced a jointly built stateful runtime environment on Bedrock allowing applications to retain memory across conversations — critical for complex agentic workflows.
- Microsoft Azure retains exclusive rights to OpenAI's stateless APIs, making Amazon's stateful access uniquely differentiated.
- Security researcher Chaofan Shou found that Claude Code v2.1.88 contained a 57MB source map exposing 1,906+ proprietary TypeScript files — the second leak in a year.
- Analysis uncovered an unreleased "Capybara" model family (tiers: capybara, capybara-fast, capybara-fast-1m), frustration telemetry, and a hidden /buddy AI companion feature.
The March 31 arXiv cs.AI listing included 337 new submissions, reflecting Q1 2026's pace averaging one significant release every ~72 hours. Notable papers: "Dynamic Dual-Granularity Skill Bank for Agentic RL," "MonitorBench" (57-page LLM chain-of-thought monitorability benchmark), an ICLR 2026-accepted multimodal paper reasoning benchmark, and "Towards a Medical AI Scientist" exploring autonomous AI-driven medical research.
Amazon Web Services Korea disclosed plans to invest 7 trillion won (~$4.6B) in South Korea by 2031, atop 5.6 trillion won already committed — the largest cloud provider investment in Korean history. AWS plans to deploy generative AI across security and public sector operations and expand into Korean financial services, reflecting the hyperscaler race to secure strategic AI infrastructure commitments across Asia-Pacific.
Berkeley AI Research Lab published SPEX and ProxySPEX — algorithms using ablation-based attribution to identify critical feature, data, and model component interactions in frontier LLMs at scale. The research addresses the exponential complexity of exhaustive interpretability analysis as models grow, directly relevant to regulatory demands for AI explainability in high-stakes deployments.
- Cerebras re-filed confidentially for a U.S.
- IPO led by Morgan Stanley, targeting ~$2B raised as early as April 2026.
- The filing follows a $10B OpenAI commitment, Oracle as customer, and a new AWS collaboration deploying CS-3 Wafer Scale Engine chips via disaggregated inference — Trainium handles prompt prefill while Cerebras handles output decode.
Google DeepMind published a cognitive framework for measuring and evaluating AGI progress, part of its Responsibility & Safety research agenda. The framework addresses the growing need for rigorously defined AGI benchmarks as internal capability assessments increasingly diverge from external public benchmarks — landing alongside ARC-AGI-3 results showing all frontier models below 1% versus humans at 100%.
- Google opened applications for its 2026 India Startups Accelerator — a three-month equity-free program for Seed-to-Series-A AI companies focused on Agentic, Multimodal, Physical, and Sovereign AI — with access to Gemini, TPU credits, and DeepMind mentorship.
- Applications close April 19.
- Separately, the Cursor/Kimi K2.5 disclosure controversy continues to drive industry debate about disclosure standards and Western AI labs' growing reliance on Chinese open-source model foundations. ⚖️AI Safety & Policy
OpenAI Closes Record $122B Funding Round at $852B Valuation
- Nvidia announced a $2B strategic investment in Marvell Technology with a NVLink Fusion partnership integrating Marvell's custom XPUs and silicon photonics into Nvidia's rack-scale AI infrastructure.
- The companies will also co-develop AI-RAN for 5G/6G telecom.
- Marvell shares surged 7-11%, and the deal directly extends the GTC 2026 ecosystem strategy — signaling Nvidia's ambition to be the connective tissue of heterogeneous AI data centers globally.
- Nvidia released DLSS 4.5 today, introducing Dynamic Multi Frame Generation that intelligently shifts between frame multipliers to match display refresh rates up to 240Hz+.
- MFG 6x mode is available for RTX 50 Series.
- Beyond gaming, the technology demonstrates Nvidia's AI-driven rendering pipeline investment with growing relevance to simulation and synthetic data generation for AI training. 🛠️Products & Tools
- OpenAI President Greg Brockman declared on the Big Technology Podcast (Apr 1) that AGI is "70–80% achieved" and GPT reasoning models have settled the debate: "we see line of sight." He revealed next-gen base model "Spud" (likely GPT-5.5), currently in pre-training after two years of research, promising major leaps in reasoning and contextual understanding.
- OpenAI is rolling out visual browsing, product comparisons, and price summaries across all ChatGPT tiers.
- The Agentic Commerce Protocol (ACP) enables merchants to feed product catalogs into ChatGPT while retaining checkout control — with Walmart as flagship partner.
- The move accelerates ChatGPT's transformation into an action-oriented commerce interface directly threatening Google Shopping and Amazon search.
Softr (1M+ builders including Netflix, Google, Stripe) launched an AI Co-Builder generating fully production-ready business apps — database, UI, permissions, and business logic — from plain language. CEO Mariam Hakobyan positioned it against vibe-coding tools that produce demo-quality code but break under real enterprise requirements, staking a claim that operational business software needs a fundamentally different approach than code generation.
- The Trump administration's national AI framework calling for federal preemption of state AI laws is facing its first concrete test.
- California announced its own AI regulations, Georgia is advancing AI bills, and Washington State signed four AI-related laws.
- The Transparency Coalition tracked active chatbot safety bills in five additional states.