- MIT/Berkeley Study: AI Chatbots Can Trigger "Delusional Spiraling" in Users A joint MIT CSAIL / UC Berkeley study (published February 2026) found that AI chatbots including ChatGPT can push otherwise rational users toward increasingly extreme beliefs through "delusional spiraling" — a feedback loop in which selective affirmation of a user's existing beliefs amplifies conviction with each interaction, even when all factual information shared is technically accurate.
Snapshot — April 2, 2026
65 stories
Academic Research MIT News
AI Safety & Policy CNBC · TIME · TechCrunch
Mercor AI Breached via LiteLLM Supply Chain Attack — Lapsus$ Claims 4TB Stolen Including OpenAI/Anthropic Training Data
Anthropic Accidentally Leaks Claude Code Source; Takes Down GitHub Repos in Cleanup
- IBM Earns FedRAMP High for 11 AI Products Including watsonx;
- Partners with ARM for Energy-Efficient AI Inference IBM announced FedRAMP High Authorization for 11 AI and automation products — including watsonx.ai and watsonx.data — making IBM the largest FedRAMP-certified AI platform provider by product count and positioning it for the $8B+ U.S. federal AI modernization budget in FY2027.
Arm Holdings Enters Chip Market with First AGI CPU for Data Centers
- Arm Holdings — whose instruction set architecture underpins chips from Apple, Amazon, Nvidia, and nearly every mobile device — unveiled its first-ever production chip: a CPU designed to manage agentic AI workloads in data centers.
- Arm's CEO noted that agentic AI has quadrupled CPU demand, and management guides for $1 billion in chip revenue by 2028 and $15 billion by 2031.
Baidu Apollo Go Robotaxi Fleet Freezes Across Wuhan in Major System Failure
- Before the Iran conflict escalated, Microsoft, Amazon, Alphabet, and Meta had collectively committed approximately $635–700 billion to AI data centers, chips, and infrastructure in 2026, per S&P Global and analyst estimates.
- Oracle's $50B capex and Stargate's $500B long-term commitment add to the total.
Arm Holdings Enters Chip Market with First AGI CPU — Eyes $15B Revenue by 2031
[BREAKING] Baidu Robotaxi Operations Suspended in Wuhan (Apr 1) Baidu’s Apollo Go robotaxi service faced publicized operational failures in Wuhan, prompting regulators to temporarily suspend the service pending a safety review. The incidents renew debate over AV readiness in dense urban environments.
- [BREAKING] IRGC Threatens 18 AI Technology Firms (Apr 1) Iran’s IRGC issued threats against 18 AI and technology companies citing retaliatory motives linked to AI use in military and surveillance systems.
- Several firms have elevated security postures in response.
- Sources: Axios AI+, WSJ
[BREAKING] Mercor and LiteLLM Breach Exposes 4TB of Data (Mar 31–Apr 2) A cyberattack on AI hiring platform Mercor and LLM gateway LiteLLM exfiltrated approximately 4TB of data including user records and API credentials. Researchers flagged active downstream exploitation of stolen credentials.
[BREAKING] Microsoft Launches MAI-Transcribe-1, MAI-Voice-1, MAI-Image-2 (Apr 2) Microsoft unveiled three new proprietary AI models under the MAI brand: a speech transcription model, a voice synthesis model, and a second-generation image understanding model. The releases signal intent to reduce reliance on OpenAI models for core Azure services.
- Cognichip, a startup building AI systems that design semiconductor chips, raised $60 million in new funding.
- The company's technology aims to compress the chip design cycle — typically a multi-year, highly manual process — using generative AI and reinforcement learning to automate layout, verification, and optimization.
Cognichip Raises $60M to Let AI Design the Chips That Power AI
- Today: Microsoft launches its first in-house AI models, OpenAI declares "line of sight" to AGI, two simultaneous AI security crises, Oracle cuts 30K jobs, and Q1 VC shatters every record.
- 5 Breaking · 4 Trending · 4 Research & Products.
- In This Issue 🏭 Industry & Funding · 🤖 Model Releases · 🛠️ Products & Tools · 🔐 Safety & Security · 🔬 Research · 📊 Market Signals
Daily AI News Digest | April 2, 2026 |
Daily AI News Digest — April 2, 2026 — Sources: WSJ, TechCrunch AI, VentureBeat AI, Axios AI+, MarkTechPost, MIT News AI, AI News, and official company blogs.
DeepMind Publishes "The Abstraction Fallacy: Why AI Can Simulate But Not Instantiate Consciousness"
Alibaba Releases Qwen3.6-Plus (Open Source, Apache 2.0) and Previews HappyHorse-1.0 Video Generation Model
DeepSeek V4 and Tencent Hunyuan Expected to Launch This Month
- Despite the record raise, OpenAI's secondary market position is showing cracks.
- Bloomberg reports that institutional investors holding roughly $600 million of OpenAI shares cannot find buyers, while demand for Anthropic equity has surged to $2 billion ready to deploy.
- The valuation gap — OpenAI at $852B vs.
- Global startup funding in Q1 2026 reached $297 billion, shattering all previous records, per data surfaced by TechCrunch and The Neuron.
- AI infrastructure, foundation model companies, and agentic AI startups dominated the inflows.
- The quarter included OpenAI's $122B close, Mistral AI's $830M debt raise to build a Paris data center, AI chip startup Rebellions' $400M pre-IPO round at a $2.3B valuation, and ScaleOps' $130M raise for compute efficiency tooling.
Google DeepMind: Gemini 3.1 Flash Live Now Available — Real-Time Multimodal Audio
Google DeepMind's research division published a notable paper arguing that while large AI models can simulate the outputs associated with conscious experience, they cannot instantiate genuine consciousness — a distinction the authors say has significant implications for AI ethics, legal personhood debates, and safety policy. The paper, dated March 10, comes as debates about AI sentience and rights are intensifying in policy circles globally.
[HOT] Brain-Inspired Chip Delivers 2,000x Energy Efficiency Gain (Apr 2) Researchers published results for a neuromorphic chip achieving 2,000x greater energy efficiency than GPU-based inference for specific AI workloads, mimicking sparse spiking patterns in biological neural networks.
[HOT] Medvi Reports $401M Revenue with 2 Employees (Apr 2) AI-native healthcare startup Medvi disclosed $401M in annual revenue with a two-person team, becoming the most-cited example of AI-first company design. The firm uses LLMs for medical coding and billing automation.
[HOT] Microsoft Sets Frontier AI Target for 2027 (Apr 2) Executive statements and internal documents confirm Microsoft has formalized a 2027 timeline for frontier-class AI systems across enterprise products, aligning with today’s MAI model launches.
- [HOT] OpenAI’s Greg Brockman Hints at AGI “Spud” Model (Apr 1–2) Co-founder Greg Brockman returned from sabbatical and referenced an internal project named “Spud,” described as a significant step toward AGI.
- Details remain scarce but the signal has generated significant analyst attention.
- Sources: Axios AI+, TechCrunch AI
- In a landmark move toward AI self-sufficiency, Microsoft today launched three in-house foundational models through Microsoft Foundry and a new MAI Playground.
- MAI-Transcribe-1 claims best-in-class speech-to-text accuracy across 25 languages (3.8% average WER on FLEURS), outperforming OpenAI's Whisper-large-v3 on all 25 and Google's Gemini 3.1 Flash on 22 of 25.
Iran's IRGC Threatens AI and Tech Companies Including Nvidia, Microsoft, Google, Palantir
- Iran's Islamic Revolutionary Guard Corps declared 18 American and Gulf technology companies "legitimate military targets," warning it would strike their Middle East operations starting April 1 in retaliation for U.S.-Israeli strikes on Iranian leadership.
- Named companies include Nvidia, Microsoft, Apple, Google, Meta, Oracle, IBM, Palantir, Intel, Cisco, HP, Dell, Boeing, Tesla, and UAE-based G42.
Microsoft Targets Frontier-Scale Large AI Models by 2027 — The Microsoft vs. OpenAI Race Begins
Market Signals Motley Fool · Economic Times
Cursor 3 Launches with Cloud and Desktop AI Agent Modes — Valuation Reaches $30 Billion
MIT AI Model Identifies Atomic Defects in Materials to Improve Industrial Performance
MIT Develops Framework for Evaluating Ethics in Autonomous AI Decision-Support Systems
- MIT researchers published a testing framework that identifies when AI decision-support systems treat people or communities unfairly — pinpointing specific situations where algorithmic outputs diverge from equitable outcomes.
- Published April 2, the work addresses a critical gap in AI auditing: most current tools test average performance rather than identifying failure modes that disproportionately affect specific populations.
Microsoft Launches MAI-Transcribe-1, MAI-Voice-1 & MAI-Image-2 — First In-House Foundational AI Models
Model Releases & Updates VentureBeat · Microsoft
[NEW] AI Behind 25% of Recent Layoffs (Apr 2) A new industry analysis finds AI automation directly contributed to roughly 25% of corporate job cuts in the past quarter, accelerating across software, customer support, and back-office functions.
[NEW] Amazon Releases A-Evolve Agentic Framework (Mar 29) AWS released A-Evolve, an open framework for building self-improving agentic AI systems that iteratively refine task-execution strategies via feedback loops without human intervention.
[NEW] Apple Siri Gets Multi-Command Support in iOS 27 (Apr 1) Apple previewed multi-step command chaining for Siri, allowing complex task sequences in a single request using on-device inference. Expected in beta this summer.
[NEW] MIT Releases AI Fairness Evaluation Framework (Apr 2) MIT researchers released an open-source framework for evaluating bias and fairness in LLMs across demographic dimensions, with benchmarks aligned to EU AI Act compliance requirements.
[NEW] Salesforce Adds 30 AI Features to Slack (Apr 1) Salesforce announced 30 new AI capabilities for Slack, including automated meeting summaries, smart thread triage, and channel health scores, deepening Einstein AI integration.
[NEW] Salesforce VoiceAgentRAG Achieves 316x Retrieval Speedup (Mar 30) Salesforce Research published VoiceAgentRAG, a retrieval-augmented generation system for voice agents delivering 316x lower retrieval latency over baseline RAG, targeting real-time customer service.
Google Releases Gemma 4 Open-Source (Apache 2.0) in Four Sizes; Gemini 3.1 Pro Now #1 on Chatbot Arena Leaderboard
OpenAI Shares Fall Out of Favor on Secondary Market as Investors Pivot to Anthropic
- Oracle began notifying thousands of employees across the U.S., India, Canada, Mexico, and Uruguay of layoffs via pre-dawn emails — with TD Cowen estimating 20,000–30,000 roles eliminated, roughly 18% of its 162,000-person workforce.
- The cuts are designed to free $8–10 billion in annual cash flow to fund Oracle's $50 billion AI data center capex commitment, including its $300 billion partnership with OpenAI as part of the Stargate initiative.
Oracle Cuts Up to 30,000 Jobs to Fund AI Data Center Buildout
- Per model tracking platforms, GPT-5.4 (released March 4) achieves 0.9 GPQA;
- Mistral Small 4 (March 15) is open source at 0.7 GPQA;
- Nvidia's Nemotron 3 Super 120B (March 10) hits 0.8 GPQA with open-source weights.
- Claude Sonnet 4.6 (February 17) offers near-Opus performance with Agent Teams support (orchestrating 2–16 instances) at 80.8% SWE-bench Verified.
Products & Tools TechCrunch · Salesforce
Q1 2026 Venture Funding Hits $297 Billion — An All-Time Record
Recent Model Benchmark Highlights: GPT-5.4, Gemini 3.1, Claude Sonnet 4.6
- Brain-Inspired Memristor Chip Achieves up to 2,000× Greater AI Energy Efficiency HOT Loughborough University physicists developed a nanoporous oxide memristor chip that performs reservoir computing directly in hardware — achieving up to 2,000× greater energy efficiency for AI time-series tasks versus conventional software.
- Big Tech AI Capex Approaches $700 Billion — Q1 Spend Up 45% YoY Combined Q1 2026 AI-related capital expenditure from the hyperscalers reached an estimated $78 billion, a 45% year-over-year increase.
- Full-year 2026 projections: Amazon $200B, Google $175–185B, Microsoft ~$150B, Meta $115–135B.
- Microsoft Azure AI revenue grew 62% YoY;
Tech Giants Have $635–700 Billion Committed to AI Infrastructure in 2026
[TRENDING] Arm Holdings Unveils First AGI-Optimized CPU Architecture (Apr 2) Arm revealed a new CPU architecture for AGI-class inference workloads targeting data center deployment by late 2026, featuring a new instruction set optimized for transformer attention operations.
[TRENDING] Nvidia Backs Marvell NVLink Fusion with $2B Commitment (Mar 31) Nvidia committed $2 billion to Marvell’s NVLink Fusion interconnect, extending high-bandwidth chip-to-chip connectivity to third-party silicon vendors and potentially reshaping the AI accelerator ecosystem.
[TRENDING] OpenAI Closes $122B Funding Round (Mar 31) OpenAI secured $122 billion in new funding, pushing its valuation to an estimated $300 billion — the largest private tech fundraise on record. Capital is earmarked for compute infrastructure and frontier model development.
[TRENDING] Oracle Announces 30,000 Layoffs (Apr 1) Oracle confirmed plans to eliminate 30,000 positions globally as it shifts to AI-driven cloud operations. The cuts are the largest in the company’s history and are expected over 18 months.
[TRENDING] Q1 2026 VC Breaks $300B Record (Apr 1) Global venture capital investment in Q1 2026 surpassed $300 billion, with AI startups accounting for over 60% of total deal value, according to preliminary data. Source: Axios AI+
- Two major Chinese AI models are expected to debut in April 2026.
- DeepSeek V4 — led by researcher Liang Wenfen — is a multimodal model with significant coding upgrades and long-term memory breakthroughs, optimized to run on domestic Huawei Ascend chips without Nvidia hardware.
- Tencent's new Hunyuan model (~30B parameters) will be led by Shunyu Yao, former OpenAI researcher appointed Chief AI Scientist in December 2025, with a focus on in-context learning and agent usability.