Back to list
Daily 2026-06-03

AI Pulse Daily | 2026-06-03

AI AgentLLMAI SafetyEmbodied AIFunding

1. Meta’s WhatsApp Business AI Agent Goes Global, Token-Based Pricing

Meta announced its customer support AI bot, “Meta Business Agent,” is now globally available for all WhatsApp Business users. The feature has been tested for nearly two years in countries like India and Mexico. Businesses will be charged based on token usage, marking another significant step in Meta’s AI monetization strategy.

AI Pulse View: With over 2 billion monthly active users, embedding AI Agents into WhatsApp dramatically expands the addressable market for enterprise AI customer service. Token-based pricing suggests Meta is building a cloud-service-like billing model for AI, creating a differentiated competitive position against OpenAI and Anthropic.

Source: TechCrunch, 2026-06-03

2. Microsoft Double Launch: Scout Personal Assistant + Project Solara Agent OS

Microsoft made two major announcements at Build. Scout is a persistent AI personal assistant built on the OpenClaw framework, deeply integrated into Microsoft 365, allowing users to name it and customize its personality. Project Solara is an Android-based operating system designed for AI Agents rather than apps, using AI to generate contextual interfaces — signaling Microsoft’s strategic pivot: “missed the app era, won’t miss the agent era.”

AI Pulse View: Microsoft’s Agent OS play is strategically significant. Project Solara attempts to redefine human-computer interaction — from “humans operating apps” to “Agents autonomously completing tasks.” If successful, this could be the biggest platform shift since smartphones. Scout and Solara together cover both personal assistant and OS layers, showing Microsoft is betting big on the AI Agent ecosystem.

Source: TechCrunch, 2026-06-02 | Ars Technica, 2026-06-02

3. OpenAI Codex Launches Six White-Collar Tools, Covering Investment Banking and Creative Design

OpenAI released six profession-specific tool plugins for the Codex app, covering data analytics, creative production, sales, product design, equity investing, and investment banking. Each tool bundles integrations, instructions, and context to enable Codex to approximate professional roles. An internal report also shows Codex usage now extends far beyond software engineering.

AI Pulse View: OpenAI is repositioning Codex from “coding assistant” to “white-collar universal AI workstation.” The six tools cover core high-value knowledge work scenarios, especially investment banking and equity roles traditionally considered hard to automate. This signals AI’s evolution from辅助 tools toward potential replacement roles in white-collar positions.

Source: TechCrunch, 2026-06-02

4. Anthropic Secretly Files for IPO, Claude Mythos Expands to Critical Infrastructure in 15 Countries

Anthropic has confidentially filed for an IPO, following a $65 billion funding round that valued the company at nearly $1 trillion. Simultaneously, Anthropic expanded its Project Glasswing security vulnerability program and Claude Mythos access to 150 organizations across 15 countries, targeting critical infrastructure in power, water, healthcare, and communications.

AI Pulse View: The $1 trillion valuation positions Anthropic for potentially the highest-valued IPO in history. More critically, Claude Mythos entering national infrastructure — power grids, water systems, healthcare — represents a breakthrough in AI’s application at national security levels, while also raising profound questions about AI dependency and systemic risk.

Source: TechCrunch, 2026-06-02

5. Google Rolls Out AI Deepfake Call Detection to Protect Android Users

Google announced a fake call detection feature for Android phones to identify phone scams using AI deepfake technology to impersonate authority figures, family members, or employers. As more people reject calls from unknown numbers, scammers are shifting to spoofing trusted numbers with AI voice cloning.

AI Pulse View: AI deepfake scams have become a global security threat. Google embedding detection into Android builds an AI security defense at the platform level. This foreshadows “AI vs. AI” as the dominant paradigm in cybersecurity — using AI to detect AI-generated content.

Source: TechCrunch, 2026-06-02 | Ars Technica, 2026-06-02

6. Trump Signs Revised AI Executive Order: Model Review Shrinks from 90 to 30 Days, Voluntary

President Trump signed a revised AI executive order requiring AI companies to voluntarily submit powerful models for government review 30 days before public release. A previous draft mandated 90 days of advance review, which was significantly reduced to voluntary after strong industry pushback.

AI Pulse View: The reduction from 90-day mandatory to 30-day voluntary reflects the AI industry’s significant lobbying power. While “voluntary” means limited enforceability, it at least establishes an institutional framework for government involvement in AI model review. The long-term challenge remains balancing innovation speed with public safety in AI governance.

Source: TechCrunch, 2026-06-02

7. Coralogix Raises $200M to Build the AI Agent Monitoring Layer

Coralogix completed a $200 million Series F round, reaching a $1.6 billion valuation, just 11 months after its $115 million Series E. The company focuses on building monitoring infrastructure for AI agents, with the round led by Advent and Canaan.

AI Pulse View: AI Agent observability is becoming a hot investment area. As enterprises deploy AI agents at scale, observability, performance monitoring, and anomaly detection become essential. Coralogix’s rapid valuation growth signals that capital markets have validated the thesis: “AI Agents need their own monitoring layer.”

Source: TechCrunch, 2026-06-03

8. Uber Caps Employee AI Spending at $1,500/Month After Burning Through Budget in 4 Months

According to Bloomberg, Uber instituted a $1,500 monthly cap per employee on AI tool usage — including Claude Code and Cursor — after exceeding its AI budget within four months of encouraging staff to use AI as much as possible. Employees can track their AI usage via an internal dashboard.

AI Pulse View: Uber’s experience reveals an overlooked reality: the cost of enterprise-scale AI tool deployment can far exceed expectations. When every engineer uses premium AI tools billed by token or subscription, total costs grow exponentially. This serves as a wake-up call for companies pushing AI adoption — fine-grained usage management and cost control mechanisms are essential.

Source: TechCrunch, 2026-06-02

9. Microsoft Open-Sources ASSERT Framework: AI Behavior Testing from Text Descriptions

Microsoft released ASSERT (Adaptive Spec-driven Scoring for Evaluation and Regression Testing), an open-source framework allowing developers to quickly create AI behavior tests through text descriptions. The framework aims to simplify the evaluation and regression testing process for AI systems.

AI Pulse View: AI system testing remains one of the industry’s hardest problems — traditional testing methods struggle with AI’s non-deterministic outputs. ASSERT’s approach of auto-generating test cases from text specifications offers a novel direction for AI behavior evaluation. As an open-source tool, it could help drive industry-wide AI testing standards.

Source: TechCrunch, 2026-06-02

10. China AI Brief: Doubao Paid Tier, ByteDance Open-Sources Bernini, Embodied AI Funding

AI Pulse View: China’s AI commercialization is accelerating — Doubao’s paid tier and Volcengine’s raised revenue targets mark the critical transition from “burning cash for users” to “monetization.” Meanwhile, embodied AI funding is heating up, with Stardust Intelligence’s 10 billion RMB valuation showing capital markets’ strong conviction in the “AI + Robotics” direction.

Sources: 36Kr, 2026-06-03 | QbitAI, 2026-06-02 | 36Kr, 2026-06-03