Meta killed its Muse Image AI feature three days after launch. Hollywood had had enough.
Meta discontinued its Muse Image AI feature after three days due to privacy concerns with the tool. The model, developed by Meta Superintelligence Labs under chief AI officer Alexandr Wang, was removed from Instagram and the Meta AI app shortly after its Tuesday launch.
Apple sues OpenAI for stealing hardware designs, alleging employees brought prototypes to “show and tell” interviews
Apple sued OpenAI, alleging the company used current and former Apple employees to steal hardware designs for AI devices. The lawsuit names OpenAI's chief hardware officer Tang Tan and former Apple engineer Chang Liu as defendants in California federal court.
US cybersecurity agency CISA had to build its incident playbook during the incident, agency reveals
A CISA contractor employee uploaded exposed passwords to a publicly accessible GitHub repository, which a GitGuardian security researcher discovered and reported in May. The agency had to develop its incident response playbook while actively managing the breach situation itself.
Apple Sues OpenAI for Trade Secret Theft
Apple sued OpenAI for trade secret theft, claiming the company encouraged Apple employees to share information about upcoming products, components, and drawings. OpenAI's hardware chief was accused of coordinating the campaign to obtain confidential Apple materials.
I built an agentic AI clone of my family to plan our summer travel
An AI system built with Claude uses multiple simulation rounds to plan family summer travel itineraries across multiple destinations with three children. The system generates detailed travel plans, though the creator notes significant limitations alongside its capabilities.
US Eases Export Curbs on UAE, Opening Door for AI Chip Sales
The Trump administration is easing export restrictions on the United Arab Emirates, permitting sales of advanced semiconductors and other technologies to support the country's AI development. The policy change allows the Gulf state broader access to chips previously subject to export controls.
Home robots already walk. 1X’s new hands try to solve the part that actually matters
1X has equipped its NEO home robot with new hands designed to handle delicate tasks like lifting wet glasses. The development addresses a critical gap where humanoid robots can walk but lack the dexterity needed for practical household use.
Shared API keys expose AI agents at 69% of enterprises, new VentureBeat research finds
Sixty-nine percent of enterprises share API keys across multiple AI agents, creating security risks where a single compromised agent gains accumulated permissions across all connected workflows. Palo Alto Networks, CrowdStrike, and Cisco have invested over 22 billion dollars in the past year on enterprise security solutions addressing this vulnerability layer.
Google will now tell you which ads were made with AI, if the advertiser admits it
Google is adding labels to advertisements that disclose whether they were created or edited with generative AI tools. The disclosure appears in the "My Ad Center" panel accessible through a three-dot menu or info icon on ads across Google Search.
News outlets ask a judge to sanction OpenAI in their landmark copyright fight
News publishers including the New York Times requested sanctions against OpenAI in federal court, alleging the company is concealing evidence relevant to their copyright infringement case. The filing claims OpenAI chose obstruction over providing datasets central to determining whether the company unlawfully used copyrighted material in training ChatGPT.
New Jersey could ban Tesla’s Robotaxi with one line about sensors
New Jersey's proposed legislation would require fully driverless vehicles to be equipped with a camera system plus two additional sensors. The incomplete bill text suggests this requirement could effectively prevent Tesla's robotaxi from operating in the state without modification.
Why you should think twice before letting an AI notetaker in your meeting
AI notetakers use speech recognition and large language models to record, transcribe, and summarize meeting conversations within seconds. Privacy advocates worry these tools store confidential information like personnel data and trade secrets in uncertain locations and create voiceprints without consent.
AI’s land rush is reaching Native American reservations
Over 100 proposed AI data center projects are being tracked on or near Native American tribal lands and rural territory. Developers are targeting these areas because tribes often possess available space, water rights, and power infrastructure, plus potential tax benefits.
Enterprises using multiple AI models are underestimating failure rates by 2.25x
Enterprises using multiple AI models underestimate failure rates by 2.25 times because they ignore the co-failure ceiling, where all models in a pool simultaneously fail on the same prompts. A study of 67 frontier models from 21 providers found that orchestration success depends not on model disagreement rates but on the percentage of prompts where every model gives wrong answers at once.
Zuckerberg confirms Meta is eyeing an AI cloud business to rent out its compute
Meta is exploring an AI cloud business to rent out its computing power to external customers. CEO Mark Zuckerberg confirmed the company believes selling access to its spare AI compute capacity makes sense.
AI companies are spending hundreds of millions on the midterms, and they want one thing
AI industry groups are spending hundreds of millions on 2026 midterm elections through super PACs. Their primary policy goal is establishing a single national AI regulatory framework instead of separate state-by-state regulations.
Palo Alto’s CEO says AI prices must fall 90% before the tech really takes off
AI token prices may need to fall by as much as 90% before businesses can deploy the technology at scale, according to Palo Alto Networks' chief executive. The CEO made this statement in response to OpenAI's claim that its new GPT-5.6 model achieved a 54% efficiency improvement.
The enterprise AI challenge nobody solves with code generation alone
Only 12 to 16 percent of organizations reach AI-driven execution despite 81 percent having detailed strategies, because generated code fails when deployed in existing enterprise environments rather than due to code quality issues. Enterprises struggle with data integration readiness, governance for AI agents executing workflows, and shifting development team roles as AI handles more coding work.
One interface isn't enough for enterprise AI
Organizations will adopt enterprise AI through multiple interfaces tailored to different business functions rather than a single unified system. Finance teams requiring strict controls, analytics groups exploring data, and customer service focusing on response times will each implement AI differently based on their distinct operational constraints.
NATO is building an AI ‘Kill Web’ to stop a Russian attack before it starts
NATO is constructing an AI network along its eastern border to detect and respond rapidly to potential Russian attacks through the Eastern Flank Deterrence Initiative. The system aims to identify threats early and enable quick counterstrikes before an attack fully materializes.
‘We cannot choose to become idiots’: a Brown professor’s proof of mass AI cheating
A Brown University economics professor observed his take-home midterm average drop from 96 to 48 when he switched the final exam from take-home to in-person format, suggesting widespread AI use during the unsupervised test. The professor publicly shared these results as evidence that most students in his class likely used artificial intelligence to complete the take-home assignment.
NHTSA demands autonomous vehicle companies fix first responder interference by end of July
Autonomous vehicles have driven into emergency scenes, blocked ambulances and firefighters, and failed to recognize flashing lights and other first responder signals. NHTSA ordered developers to submit solutions for this interference issue by the end of July.
SpaceX's Grok 4.5 launches at half the price of rivals — here's why that could rattle Anthropic and OpenAI
Elon Musk's SpaceX released Grok 4.5, trained specifically for coding and autonomous agents using its $60 billion Cursor acquisition, priced at $2 per million input tokens and $6 per million output tokens. The model uses half as many tokens per task as comparable models and costs less than half as much as rival premium offerings from Anthropic and OpenAI.
AI money watch: five funding rounds that matter today
A quantum computing company raised $300 million while an AI-agent startup achieved a billion-dollar valuation. European startups focused on energy also received funding within a 24-hour period.
Irish drone delivery startup Manna is building a 1,000-person US factory in Tulsa to take on Zipline and Wing
Manna Aero is building a manufacturing facility in Tulsa, Oklahoma that will employ approximately 1,000 people, with production starting in about one year. The Irish drone delivery company raised $50 million in venture capital to fund this US expansion.
Google’s SynthID watermark just debunked its first high-profile deepfake: a hoax image of Mitch McConnell
Google's SynthID watermarking system identified an AI-generated image of Senator Mitch McConnell circulating on Reddit and X as fake, marking its first high-profile real-world application. Snopes used the technology to debunk the viral image, which depicted McConnell in a hospital bed.
With EU backing, QuantumDiamonds aims to speed up chip manufacturing
QuantumDiamonds, a German startup, received EU backing to develop chip inspection technology using a novel approach. The company operates within the European Chips Act framework, which provides state subsidies to strengthen the semiconductor industry.
HubSpot tried to feed its AI with customer data. The revolt took four days
HubSpot reversed a policy change that would have used customer data for AI lead-finding features after four days of backlash. The company had automatically opted customers into data pooling on July 1 before scrapping the plan.
Squeezed on land, Samsung wants to put data centres out to sea
Samsung Heavy Industries plans to launch its first floating data centre by 2028 using a purpose-built barge parked near the coast. The facility will be specifically designed rather than converted from an existing ship.
Nvidia teams up with chip rival d-Matrix instead of fighting it
Nvidia combined its hardware with inference chips from startup d-Matrix to create a joint system for running AI models. Parasail will be the first customer to receive this combined system.