Samwise Tech/AI/Robotics Newsletter
Monday, June 29, 2026
OpenAI limits GPT-5.6 rollout after government request, says restrictions shouldn’t be the norm
OpenAI agreed to limit the rollout of GPT-5.6 following a direct request from the U.S. government, restricting initial access rather than launching broadly as originally planned. The company stated it does not view such government-prompted restrictions as standard practice and cautioned against them becoming the norm for AI development. The move is among the most explicit public disclosures of a leading AI lab accepting a government request to slow a specific model release. The episode underscores growing federal scrutiny over when and how frontier AI systems reach consumers and enterprises.
Source: TechCrunch Share ↗ ✉︎ Email 💬 Text
Asian AI startups launch Mythos-like models as Anthropic’s export ban drags on
Asian AI startups are launching models they describe as comparable to Anthropic’s Mythos-class systems, filling a market gap created by ongoing U.S. export restrictions that limit Anthropic’s ability to serve customers in certain overseas markets. The models are being actively positioned as domestic alternatives for enterprises unable to access American frontier AI. The situation illustrates a central tension in U.S. AI export policy: restrictions intended to contain advanced capabilities may instead accelerate independent development by foreign competitors, potentially expanding access to powerful AI systems beyond U.S. oversight or safety frameworks.
Source: TechCrunch Share ↗ ✉︎ Email 💬 Text
General Intuition raises $320M to use video game data to train robots
General Intuition has raised $320 million to develop robots trained primarily on video game data, arguing that virtual environments offer a more diverse and scalable training ground than physical datasets alone. The startup uses gameplay footage and physics simulations to expose robotic systems to a wide range of scenarios, including situations too dangerous or expensive to stage in the real world. The approach, known as sim-to-real transfer, is gaining commercial traction as robotics companies look for faster, more cost-effective methods to teach machines physical manipulation, navigation, and dexterous task execution.
Source: The Robot Report Share ↗ ✉︎ Email 💬 Text
Why Wall Street thinks US memory maker Micron is the next Nvidia
Wall Street analysts are increasingly positioning Micron Technology as a primary beneficiary of the AI hardware boom, drawing comparisons to Nvidia’s dominant decade. The investment case centres on high-bandwidth memory: as AI inference workloads scale, demand for the specialized memory chips Micron manufactures is rising rapidly, and supply has not kept pace. The article examines why analysts believe Micron’s manufacturing capabilities, long-term contracts with major cloud customers, and strategic timing in the HBM market make it well-positioned to capture an outsize share of AI infrastructure spending in the coming years.
Source: TechCrunch Share ↗ ✉︎ Email 💬 Text
AGIBOT produces 15,000th robot, marking a milestone in embodied AI deployment
Chinese robotics company AGIBOT has produced its 15,000th robot, marking a significant milestone in the scale-up of embodied AI at a manufacturing level. The achievement reflects rapid capacity expansion across China’s robotics sector, where multiple firms are racing to scale production of physical AI robots for industrial and commercial use. The output milestone signals that AGIBOT has moved decisively past the prototype and pilot phase into sustained volume manufacturing — a transition analysts consider essential before embodied AI systems can achieve the unit cost reductions needed for mainstream enterprise and consumer deployment.
Source: The Robot Report Share ↗ ✉︎ Email 💬 Text
Russian hackers were behind $2.5 billion hack of Jaguar Land Rover: Report
Russian hackers were responsible for the $2.5 billion cyberattack on Jaguar Land Rover, according to a new report — placing it among the most financially damaging automotive cybersecurity incidents ever recorded. The attribution has not been officially confirmed by JLR, and the specific Russian threat group behind the breach has not been named publicly. The sheer scale of the financial damage illustrates how exposed major manufacturers have become as vehicle systems, supply chains, and corporate infrastructure grow increasingly software-driven and network-dependent. Attribution of cyber incidents to nation-states remains politically sensitive and technically contested.
Source: TechCrunch Share ↗ ✉︎ Email 💬 Text
Rippling Data Cloud: Parker Conrad knows which employees are worth their AI spend
Rippling Data Cloud is a new platform that gives employers detailed visibility into how their workers are using AI tools and whether that spending is generating measurable productivity returns. CEO Parker Conrad argues that organizations need the same granular workforce analytics applied to AI expenditure that they already use for other operational costs. The platform draws on Rippling’s existing HR, payroll, and IT management suite to surface data on software usage patterns, productivity signals, and per-employee costs — helping businesses identify which employees are delivering a return on their AI spending.
Source: TechCrunch Share ↗ ✉︎ Email 💬 Text
Tech Pulse
Top Frontier Models: GPT-5.4 Pro (94.6%) | Gemini 3.1 Pro (94.1%) | GPT-5.5 (94.0%)
Top Open Source Models: Llama 4 Maverick (85.5%) | Qwen 3 235B (85.7%) | DeepSeek R1 (97.3% MATH)
Top Small Models (15–50B): Phi-4 14B (84.8%) | Mistral Small 3.1 (79.0%) | Gemma 3 27B (78.6%)
Top Edge Models (0–15B): IBM Granite 4.1 8B (87.2% HE) | Phi-4 Mini 3.8B (67.3%) | Gemma 3 4B (59.6%)
AI Leaders: NVIDIA $5.23T | Alphabet $4.63T | Microsoft $3.11T
Robotics Leaders: Intuitive Surgical $160B | FANUC $38.9B | ABB Robotics $3.5B est.
Curated by JD · samwise.agency
Leave a Reply
You must be logged in to post a comment.