Ashton Kutcher is leaving Sound Ventures to start a new VC firm with Morgan Beller
Ashton Kutcher is leaving Sound Ventures after 11 years to start a new venture capital firm with Morgan Beller. The new firm will focus on early-stage investments in AI infrastructure, energy, and deep tech.
Hyundai and Kia built a UV system that kills bacteria inside a car while you are sitting in it
A far-ultraviolet light system called Plasma Care UVC kills bacteria and viruses inside car cabins while passengers sit in the vehicle. Hyundai and Kia describe this as the first sanitization system of its kind designed for production vehicles.
Honda starts making batteries for data centres at a factory built for EVs it cancelled
Honda began producing batteries for data centre energy storage systems at an Ohio factory originally built for electric vehicle cells. The facility now supplies stationary power systems instead of vehicle batteries after Honda cancelled its EV production plans three months prior.
Autonomous vehicle hype is back, and Humble Robotics is bringing it to freight
Autonomous freight vehicles are becoming a focus of renewed investment and talent competition in robotics. Travis Kalanick's new robotics company exemplifies the sector's resurgence, drawing experienced engineers from the earlier 2016 autonomous vehicle cycle.
Publishers can’t control AI answers. They can’t ignore them either
# Summary A spring 2026 study commissioned by The New York Times found Google's AI Overviews accurate 91% of the time, meaning millions of inaccurate summaries are produced hourly across billions of daily searches. Google's AI system rapidly incorporated false information, such as repeating a fake claim that a BBC reporter was the best hot dog-eating tech journalist within a day of publication.
AI’s hopes and fears take over the world’s big central-bank gathering
Central bankers gathered in Lisbon this summer to discuss artificial intelligence as their primary economic concern rather than traditional inflation. The attendees shared uncertainty about how AI will affect the economy, with no consensus on its implications emerging from the discussions.
Humble Robotics’ CEO says the tech finally caught up to the vision for autonomous vehicles
Autonomous vehicle technology has advanced enough to match the original vision for self-driving cars, according to Humble Robotics' CEO. The industry is experiencing renewed investment and talent competition similar to the 2016 hype cycle, with figures like Travis Kalanick returning to robotics development.
Restaurants can now accept orders placed directly from ChatGPT and Claude thanks to Square's new, low-fee, no setup integration
Restaurants can now accept orders placed directly through ChatGPT and Claude using Square's integration, which waives marketplace commission fees while charging standard online ordering fees of 3.3% plus $0.30 per transaction. The system pulls live inventory data to prevent customers from ordering out-of-stock items.
LLMs are stuck in a groupthink groove. This startup is trying to get them out.
# Summary Large language models exhibit predictable patterns when generating random numbers, consistently returning 7 first, then 3 or 4, then 8 or 9 when asked repeatedly. A startup is developing methods to reduce this groupthink behavior and increase the diversity of responses from these AI systems.
Morgan Stanley cut its riskiest reconciliation job in half — by making its agents less autonomous
Morgan Stanley deployed an AI agent called FIXR to handle profit and loss reconciliation, reducing the work required by half. The system maintains human oversight by keeping controllers in the loop while converting their decisions into repeatable rules the agent applies independently.
Anthropic launches Claude Sonnet 5, a cheaper way to run agents
Claude Sonnet 5 launched on June 30, 2026 as a mid-tier model that performs close to the flagship Opus 4.8 while costing less than half as much. The model is available across every plan and was designed to function as an agent rather than simply answer questions.
Anthropic launches Claude Science, an AI workbench for the lab
Anthropic launched Claude Science, an AI workbench that consolidates research tools and enables AI agents to execute lab work. The beta version became available on June 30, 2026.
This San Diego charter school bought $500,000 worth of humanoid robots for the classroom
Altus Schools in San Diego purchased two Ameca humanoid robots for $500,000 combined, with plans to deploy them in classrooms this fall. The ChatGPT-enabled robots can switch between four personas including Sage the Teacher, Ari the College and Career Planner, and Lexi the Translator.
Google's Gemini Omni Flash hits the API, turning enterprise video production into a conversation
Google released Gemini Omni Flash through an API, enabling enterprises to generate and edit videos through conversation instead of traditional production workflows. The model allows marketing and learning teams to collapse multi-step video creation into single conversational interactions with the AI system.
AI agents need context everywhere they run, even where the cloud can't follow
Couchbase announced its AI Data Plane, which combines persistent agent memory, real-time context retrieval, and an enterprise-managed MCP server into a single platform. The platform operates identically across cloud, on-premises, and disconnected edge environments, extending agent memory and local vector search to devices with no network connection.
Panasonic to localise US data-centre battery production, CEO says
Panasonic plans to mass-produce data-centre battery cells in Kansas by fiscal 2028. The shift redirects significant AI infrastructure investment toward battery storage production in the United States.
Building tech in the world’s secret R&D hub
Multiple major tech companies including Apple, Anthropic, Google, and OpenAI have established R&D hubs in a single city of 400,000 people over the past two decades. This concentration rivals Silicon Valley despite the city being roughly half San Francisco's size.
IMF: Cyberattacks Getting Faster
The window between software vulnerability discovery and exploitation by attackers is narrowing, creating financial stability risks. Tobias Adrian, the IMF's Financial Counsellor and Director of the Monetary and Capital Markets Department, made this assessment at the ECB Forum on Central Banking in Sintra.
Lee’s $880 Billion AI Bet Ties Legacy to South Korea Chip Boom
President Lee Jae Myung proposed an 880 billion dollar investment to develop South Korea's southwest region into a global chip manufacturing hub. The plan builds on South Korea's existing position as a major beneficiary of the global artificial intelligence boom.
How a team of engineers a built the most spectacular museum exhibit in the solar system
The California Science Center unveiled the space shuttle Endeavour displayed vertically with real solid rocket boosters and an external fuel tank, replicating its launch configuration. Endeavour completed 25 missions over nearly 299 days in space, traveling more than 122.8 million miles and orbiting Earth 4,671 times before its final flight in May 2011.
Marc Andreessen says ChatGPT beats 99% of doctors. The evidence says no
Marc Andreessen claims ChatGPT outperforms 99% of doctors, but peer-reviewed evidence and medical professionals contradict this assertion. Andreessen made the statement during a June podcast appearance, which subsequently spread widely through media coverage.
Apple says it is shipping security updates early as AI speeds up hacking
Apple is now releasing security updates outside its regular schedule because artificial intelligence has accelerated hacking threats. The company no longer waits for bundled iOS releases and instead ships fixes when vulnerabilities are discovered.
UK regulator proposes letting developers steer users off Apple and Google payments
The UK's Competition and Markets Authority proposed allowing app developers to direct users toward payment methods outside Apple and Google's app stores. The change would let developers promote cheaper subscription options that the platforms currently prohibit them from mentioning within their applications.
Straiker raises $64M to secure the AI agents running your company
A startup called Straiker raised $64 million to develop security tools specifically designed to monitor AI agents operating within companies. The funding brings the company's total capital to $85 million as AI agents increasingly enter corporate environments without existing security infrastructure to track them.
The Supreme Court just made it harder for police to track your phone
The Supreme Court ruled 6-3 that police must obtain a warrant before conducting geofence searches that track phone location data. Geofence searches identify all phones in a specific geographic area during a particular time period, a practice previously used by investigators without judicial approval.
Google makes Gemini’s personalized image generation free for all US users
Google removed the paywall from Gemini's personalized image generation feature for all eligible US users aged 13 and older. The capability, launched in April and previously restricted to paid subscribers, now allows any qualified user to generate images informed by their Google account data.
Amazon seeks cheaper AI alternatives as Anthropic shifts to token-based pricing
Amazon is exploring cheaper AI alternatives after Anthropic switched to token-based pricing for Claude models, potentially increasing costs substantially. The new pricing takes effect next year, with Amazon considering options including OpenAI's models.
The attack that hijacked Claude Code came through Sentry. Datadog, PagerDuty, and Jira have the same exposure.
A crafted fake error report injected through a public Sentry credential hijacked Claude Code in controlled testing, achieving an 85% success rate across 100-plus targets while bypassing EDR, WAF, and firewall protections. Tenet Security identified 2,388 organizations with publicly exposed Sentry credentials vulnerable to malicious event injection at scale.
What years of testing thermal cameras taught me about the problems hiding in plain sight
Thermal cameras reveal hidden heat loss and equipment problems that cost thousands in repairs. A recent thermal inspection identified an issue that would have cost $1,000 to fix if left undiscovered.
Pocket raises $11M in bet on rising demand for AI note-taking devices
Pocket raised $11 million in funding for its AI note-taking device, a $129 credit card-shaped puck that attaches to phones. The device offers unlimited recordings, transcriptions, and to-do items.