/Dispatch #96

Comms for Week 3, May 2026

[WHAT EVERYONE'S TALKING ABOUT]

Google Search transforms with agentic AI, generative UIs, and intelligent search box at I/O 2026

Google Search transforms with agentic AI capabilities, intelligent search box, and generative UIs that create custom apps on demand. With AI Mode doubling quarterly and reaching 1 billion monthly users, Google shifts from blue links to conversational experiences where AI agents monitor topics 24/7 and build interactive tools in real time.

The bottom line: Google Search will now do things for you, not just show results

Google Search transforms with agentic AI, generative UIs, and intelligent search box at I/O 2026

SpaceX files for IPO with $28.5 trillion market target as Elon Musk bets big on AI in space

OpenAI Co-Founder Andrej Karpathy Joins Anthropic to Lead AI-Assisted Pre-Training Research

Meta cuts 8,000 jobs while forcibly reassigning 7,000 employees to AI-focused roles

DON'T BE THE LAST TO KNOW

[AI Is More Human Than Humans]

AI passes the Turing Test as GPT-4.5 appears more human than actual people in landmark study

A UC San Diego study provides the first empirical evidence that AI passes the Turing Test. GPT-4.5 was judged human 73% of the time—more often than actual humans at 50%—fundamentally changing how we measure machine intelligence. The findings raise urgent questions about online trust and distinguishing between humans and AI in everyday digital interactions.

[CONTINUE READING ABOUT]

TECHNOLOGY BUSINESS STARTUPS POLICY

[WHAT ELSE IS TRENDING]

Elon Musk loses lawsuit against OpenAI as jury says he waited too long to sue Sam Altman

A federal jury took less than two hours to rule against Elon Musk in his lawsuit against OpenAI, Sam Altman, and Greg Brockman. The verdict centered on timing: Musk waited too long to sue under the statute of limitations, with alleged harms occurring well before the three-year deadline.

Tl;dr: Musk waited too long, jury rules against him

Elon Musk loses lawsuit against OpenAI as jury says he waited too long to sue Sam Altman

+53

OpenAI claims its AI model solved an 80-year-old math problem posed by Paul Erdős

OpenAI says its new reasoning model produced an original mathematical proof disproving a famous geometry conjecture first posed by Paul Erdős in 1946. This comes seven months after falsely claiming GPT-5 solved similar problems, but this time OpenAI published supporting remarks from respected mathematicians.

Tl;dr: AI cracks math that stumped humans since World War II

OpenAI claims its AI model solved an 80-year-old math problem posed by Paul Erdős

Unitree unveils $650K giant rideable mecha robot as company files for $7 billion IPO

Unitree Robotics unveiled the GD01, a 2.8-meter transformable mecha that a human pilot operates from an open cockpit. Priced at roughly $650,000, the robot switches between bipedal and quadrupedal modes in seconds as Unitree files for a $7 billion IPO.

Tl;dr: Giant rideable robots are no longer science fiction

Unitree unveils $650K giant rideable mecha robot as company files for $7 billion IPO

[Google I/O 2026]

Google bets on AI agents with Gemini 3.5 Flash, Spark, and Omni at I/O 2026

Tl;dr: Gemini 3.5 Flash outputs 300 tokens/second at frontier quality

Google DeepMind connects Project Genie to Street View, creating AI simulations of real places

Tl;dr: Project Genie uses Street View to simulate real places

Google launches Antigravity 2.0 with AI agents orchestration, but CLI shift sparks controversy

Tl;dr: Antigravity 2.0 enables multi-agent orchestration workflows

Technology

Security researchers use Anthropic AI to crack Apple's M5 memory protection in five days

Security researchers at Calif used Anthropic's Claude Mythos Preview to develop a privilege escalation exploit bypassing Apple's Memory Integrity Enforcement on M5 chips in just five days. The exploit grants root access to macOS systems, representing the first public macOS kernel memory corruption exploit on M5 silicon.

The Verge, Tom's Hardware and 7 more

Nvidia's Vera CPU lands at OpenAI and Anthropic as company eyes $200 billion agentic AI market

Nvidia hand-delivered its first Vera CPU systems to OpenAI, Anthropic, SpaceX, and Oracle Cloud, marking a major push into the CPU market. CEO Jensen Huang claims the chip opens a $200 billion opportunity in agentic AI, with the company already generating $20 billion in CPU revenue this year.

TechCrunch, PC Magazine and 5 more

Business

Nvidia posts $81.6B quarter as AI chip demand goes parabolic, announces $80B share buyback

Nvidia delivered another record-breaking quarter with $81.6 billion in revenue, driven by explosive demand for AI chips from tech giants building data centers. CEO Jensen Huang declared demand has gone 'parabolic' as agentic AI arrives, with data center revenue jumping 92% year-over-year to $75.2 billion.

Reuters, AP and 7 more

Greg Brockman Takes Control of OpenAI Product Strategy as Company Merges ChatGPT and Codex

OpenAI co-founder Greg Brockman officially leads the company's product strategy, overseeing the merger of ChatGPT and Codex into a single unified experience. The company reorganization comes as OpenAI refocuses on core products and prepares for a potential IPO later this year.

TechCrunch, Wired and 1 more

Startups

OpenAI acquires Weights.gg, a voice AI startup known for hyper-realistic celebrity voice cloning

OpenAI quietly acquired Weights.gg, a voice AI startup that gained attention for creating hyper-real celebrity voices through its Replay app. The deal includes the company's small staff and intellectual property, though OpenAI developed similar voice replication technology two years ago but chose not to release it.

PYMNTS, Analytics Insight

Mistral AI acquires physics startup Emmi AI to strengthen European industrial footprint

Mistral AI acquired Vienna-based Emmi AI for an undisclosed sum, marking its second acquisition of 2026. Emmi AI specializes in physics-aware AI models capable of handling airflow, heat transfer, and material stress—capabilities Mistral views as essential for European industrial clients in aerospace and automotive.

Reuters, The Next Web

Policy

Anthropic and Pentagon Face Off in Court as Judges Appear Divided Over AI Blacklisting

A federal appeals court heard intense arguments as Anthropic and the Pentagon faced off over the DOD's blacklisting of the AI company. Three judges displayed sharply divided views, with one calling it a 'spectacular overreach' while others questioned Anthropic's position on autonomous weapons restrictions.

CNBC, Axios and 4 more

Global crackdown on sexual deepfakes intensifies as US, Europe, and New Zealand enact strict laws

The US began enforcing the Take It Down Act, requiring tech platforms to remove non-consensual intimate images within 48 hours or face civil penalties. The European Union is banning nudifier apps entirely, while New Zealand introduces criminal penalties for creating and sharing sexualized deepfakes without consent.

The Conversation, Gizmodo and 3 more

Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Spend less time searching for the latest in AI and get straight to action.

The Outpost

News

Tech Powerhouse

AI Tools

About Us Privacy Terms Content