Samwise Tech/AI/Robotics Newsletter
Sunday, July 12, 2026
Apple Sues OpenAI Over Alleged Trade Secret Theft, Naming Over 400 Former Employees
Apple has filed a lawsuit against OpenAI alleging systematic trade secret theft, claiming the AI company deliberately recruited more than 400 Apple employees to extract confidential technical knowledge. The suit names former Apple design executives Tang Tan and Chang Liu as central figures in an alleged coordinated scheme to transfer Apple’s proprietary hardware and software expertise into OpenAI’s operations. Apple contends the recruitment drive stripped key engineering teams of specialized knowledge spanning chip design and on-device machine learning. The lawsuit marks a dramatic escalation of tensions between two Silicon Valley giants now in direct competition over AI product development and deep technical talent.
Sources: TechCrunch Share ↗ ✉︎ Email 💬 Text
Anthropic, OpenAI and SpaceX Combined Value Now Exceeds All Tech Exits of the Past 25 Years
A new analysis finds that Anthropic, OpenAI, and SpaceX have collectively accumulated more private market value than all technology company exits — IPOs and acquisitions combined — over the prior 25 years. SpaceX alone reached a valuation of $1.77 trillion, while OpenAI and Anthropic have each secured enormous private valuations in recent rounds. The combined figure surpasses $4 trillion, dwarfing the aggregate exit value generated by the entire tech sector since the late 1990s. The data point reflects a structural shift in where capital concentrates: increasingly in private AI and deep-tech companies that can sustain stratospheric valuations without accessing public markets.
Sources: TechCrunch Share ↗ ✉︎ Email 💬 Text
New York Times Accuses OpenAI of Hiding Evidence in ChatGPT Copyright Lawsuit
The New York Times has accused OpenAI of deliberately concealing evidence in ongoing copyright litigation over the company’s alleged unauthorized use of published journalism. Court filings revealed that OpenAI failed to preserve server logs that could have illuminated how ChatGPT was trained on Times articles and documents. The accusations centre on a deposition involving OpenAI’s general counsel, with the Times arguing the company’s conduct amounted to wilful destruction of evidence material to the copyright claim. If substantiated, the allegations could expose OpenAI to severe penalties including adverse inference instructions — allowing a jury to presume the destroyed logs would have proven infringement.
Sources: TechCrunch Share ↗ ✉︎ Email 💬 Text
Paris AI Voice Startup Gradium Raises $100M Seed Round Backed by Nvidia
Paris-based AI voice startup Gradium has secured $100 million in seed funding — an unusually large early-stage raise — with backing from Nvidia and other strategic investors. Gradium is a spinout from Kyutai, the French AI research lab known for its Moshi real-time voice work, and counts Renault among its early enterprise customers. The round underscores Europe’s growing AI investment momentum and Nvidia’s strategy of backing startups that build on its compute ecosystem. Gradium’s founding team brings expertise in low-latency speech synthesis and real-time dialogue, targeting automotive and industrial use cases where voice interaction can replace screens and keyboards across factory-floor environments.
Sources: TechCrunch Share ↗ ✉︎ Email 💬 Text
“Slopsquatting”: AI Hallucinated Package Names Are Becoming a Software Supply Chain Threat
Security researchers have documented a new class of software supply chain attack enabled by AI coding tools: “slopsquatting.” The threat exploits AI code assistants’ tendency to hallucinate non-existent package names at a rate of 19.7%, returning the same phantom identifiers consistently across repeated queries. Attackers who register these hallucinated names on PyPI, npm, or other repositories can silently intercept developer installations. Unlike typosquatting — which relies on human typing errors — slopsquatting scales through AI-generated code that developers may not carefully audit before execution. The researchers found the hallucination rate persistent across multiple frontier models, raising urgent questions about AI code assistant reliability in production pipelines.
Sources: VentureBeat Share ↗ ✉︎ Email 💬 Text
Even Realities’ G2 Smart Glasses Bet That No Camera Is the Right Call for the Enterprise
Smart glasses startup Even Realities has unveiled the G2, a 35-gram device that deliberately omits a camera — a striking reversal from competitors like Meta and Apple that have centred AI glasses around image capture. The G2 focuses instead on a 1200-nit display bright enough for outdoor use and a companion R1 ring controller for discreet hands-free interaction. The company argues enterprise buyers are more concerned about recording employees and customers than ambient intelligence features, and that privacy-first hardware unlocks corporate and medical deployments otherwise blocked by policy. The glasses support third-party AI integrations and are positioned as a productivity tool rather than a consumer lifestyle device.
Sources: TechCrunch Share ↗ ✉︎ Email 💬 Text
Reed Jobs’ Yosemite Fund Is Using AI to Accelerate Cancer Drug Discovery
Reed Jobs, son of Steve Jobs, has spoken candidly about the mission driving Yosemite, the oncology-focused venture fund he quietly launched in 2023. Rather than trade on his surname, Jobs says his aim is identifying academic research that AI can transform into viable cancer therapies. Yosemite takes early bets on university science that would otherwise stall in labs, then applies AI tools to drug discovery, clinical trial design, and patient matching. Jobs argued that AI’s capacity to analyse molecular data at scale is a genuine inflection point for oncology — not incremental progress, but a fundamental change in how quickly new treatments can be identified, validated, and brought to patients.
Sources: TechCrunch Share ↗ ✉︎ Email 💬 Text
Tech Pulse
Top Frontier Models: Claude Mythos 5 (80.3% SWE-Pro) | GPT-5.6 Sol (80.0%) | Claude Fable 5 (77.2%)
Top Open Source Models: DeepSeek V4 Pro (80.6%) | Z.ai GLM-5.2 (74.4%) | Llama 4 Maverick (73.8%)
Top Small Models (15–50B): Qwen 3.5-27B (82.5%) | Llama 4 Scout 17B (73.0%) | Mistral Large 3 (70.4%)
Top Edge Models (0–15B): Phi-4 14B (84.8%) | Qwen 3.5-9B (82.5%) | Gemma 3 4B (79.2%)
AI Leaders: NVIDIA $5.1T | Alphabet $4.3T | Microsoft $2.83T
Robotics Leaders: Intuitive Surgical $175B | Fanuc ~$43B | ABB Robotics $5.4B (est.)
Curated by JD · samwise.agency