Samwise TAIR Newsletter — Sunday, April 12, 2026

HARDWARE

Snap and Qualcomm Partner to Power Consumer AR Glasses Launching in 2026

On April 10, 2026, Snap’s wholly owned subsidiary Specs Inc. and Qualcomm Technologies announced a multi-year strategic agreement to power upcoming consumer AR glasses with Qualcomm’s Snapdragon system-on-chip. The partnership, which Snap describes as expanding a track record stretching back through multiple Spectacles generations, targets on-device AI processing, advanced graphics, and multi-user shared digital experiences. Specs are standalone, see-through glasses designed to overlay digital content on the physical world. Snap established Specs Inc. as a standalone unit on January 28, 2026, to give the business more operational focus and partnership flexibility. A consumer launch is scheduled for later in 2026.

Sources: TechCrunch

SECURITYAI

RSAC 2026: Five Vendors Ship Agent Identity Frameworks, Three Critical Security Gaps Remain

At RSAC 2026, five major cybersecurity vendors — CrowdStrike, Cisco, Palo Alto Networks, Microsoft, and Cato CTRL — each shipped new agent identity frameworks designed to govern autonomous AI systems operating within enterprise environments. Despite the flurry of vendor activity, three critical gaps survived all five launches: no framework can detect an agent rewriting its own security policy mid-task, track delegation chains when one agent hands work to another, or verify that a decommissioned agent holds zero residual credentials. Separately, CrowdStrike, Cisco, and Palo Alto Networks launched agentic SOC tools without providing the agent behavioral baseline security teams need to set effective policy.

Sources: VentureBeat

AISOFTWARE

Meta Debuts Muse Spark, Its First Proprietary AI Model Under Alexandr Wang

Meta debuted Muse Spark on April 8, 2026, the first model released by its newly formed Meta Superintelligence Labs under CEO Mark Zuckerberg and head Alexandr Wang. Code-named Avocado and built over nine months, Muse Spark accepts voice, text, and image inputs and operates in multiple modes: a fast conversational layer, a “shopping mode” blending language models with user behavior data, and a “Contemplating mode” that runs parallel reasoning agents simultaneously. Unlike Meta’s previous Llama family of open-weight models, Muse Spark launches as a proprietary system, initially powering the Meta AI app and website before rolling out to WhatsApp, Instagram, Facebook, and Meta’s AI glasses in coming weeks.

Sources: TechCrunch

AISECURITY

Anthropic Launches Project Glasswing, Deploying Claude Mythos on Defensive Cybersecurity

On April 7, 2026, Anthropic launched Project Glasswing, a cybersecurity initiative uniting twelve organizations — AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks — to defend critical open-source software using AI. The initiative was prompted by Claude Mythos Preview, which achieved 83.1% on CyberGym vulnerability benchmarks versus 66.6% for Claude Opus 4.6. Mythos found thousands of zero-day flaws in major operating systems, identified a 27-year-old OpenBSD bug, and surfaced a 16-year-old FFmpeg vulnerability missed across five million automated test attempts. Anthropic committed $100 million in model credits and $4 million in grants to open-source security foundations.

Sources: Anthropic

AIINDUSTRY

Perplexity Brings 20-Model ‘Computer’ AI Agent to Enterprise, Taking Aim at Microsoft and Salesforce

Perplexity launched Computer for Enterprise at its Ask 2026 developer conference, expanding the multi-model AI agent — which coordinates up to 20 AI models and carries a $200-per-month consumer price — into the enterprise market. The launch adds Slack integration and Snowflake data connectors, enabling employees to invoke the agent directly in Slack threads and continue conversations across Perplexity’s web and mobile interfaces. The company, valued at $20 billion, positions Computer as a direct competitor to Microsoft Copilot and Salesforce, targeting enterprise software workflows previously requiring dedicated teams. Consumer users had already demonstrated the agent building Bloomberg Terminal-style financial dashboards and replacing marketing tool stacks within a single weekend.

Sources: VentureBeat

REGULATIONAI

OpenAI Proposes Robot Taxes, Public Wealth Funds, and Four-Day Workweek for the Intelligence Age

On April 6, 2026, OpenAI released labor and economic policy proposals outlining how governments should manage wealth distribution in what it calls the “intelligence age.” The document advocates for public wealth funds financed by AI-generated profits to ensure broad societal participation in technology gains, robot taxes to raise revenue as automation displaces workers, and federal subsidies for a four-day workweek without wage reduction. OpenAI also proposed expanding the Earned Income Tax Credit and increasing investment in education and healthcare infrastructure. The policy paper frames these measures as necessary to prevent economic disruption as AI systems assume an increasingly large share of the labor market.

Sources: TechCrunch

ROBOTICSINDUSTRY

Japan Pushes Physical AI From Lab to Shop Floor, Eyes 30% of Global Market by 2040

Japan is moving physical AI from experimental pilots into commercial deployment at scale, driven by some of the world’s most acute labor shortages. The Ministry of Economy, Trade and Industry has set a target of capturing 30% of the global physical AI market by 2040, establishing domestic manufacturing ecosystems for robot systems capable of operating in unstructured environments. Startups and manufacturers in logistics, elderly care, and food service are integrating autonomous robots using reinforcement learning and vision-language-action models to perform tasks previously relying on human labor. Industry observers say Japan’s demographic constraints are compressing the typical commercial adoption timeline for physical AI by several years.

Sources: TechCrunch

Tech Pulse

Top Models (SWE-Bench): Claude Opus 4.6 (72.5%)  |  Gemini 2.5 Pro (63.8%)  |  GPT-4.1 (54.6%)

Market: NVIDIA $3.6T  |  Apple $3.2T  |  Microsoft $2.9T