Edition: July 16, 2026

Apple is suing OpenAI for allegedly stealing confidential hardware secrets, while OpenAI employees have started a super PAC to advocate for stronger AI guardrails. New York has also passed a statewide data center moratorium, sparking debate about AI regulation. Meanwhile, a cyclosporiasis outbreak is spreading across the US.

Anthropic, a nearly $1 trillion-valued AI startup, is pushing states to adopt tougher regulations on AI, citing the need for more stringent safety measures as AI capabilities advance. The company has supported laws in California, New York, Illinois, and Massachusetts, and is working to address catastrophic AI risks. Anthropic's pro-regulation stance has been met with both praise and criticism, with some accusing the company of trying to stifle smaller startups.

Google Vids now allows users to create custom digital avatars and AI videos using a selfie, voice recording, and written prompts. The update brings Google's multi-modal AI model Gemini Omni to Vids, enabling features like background swaps and effects. This expansion positions Google Vids as an all-in-one video creation platform beyond its original workplace presentation tool role.

Roblox has launched a new AI-powered game creation feature called 'Build' that allows users to design games from their mobile devices using simple text prompts. The feature, set to enter public alpha testing on July 28, aims to make game development more accessible but has raised concerns about the potential for low-quality content. Roblox plans to address this by ranking AI-generated games based on player retention.

The EU has ordered Google to share search data with competitors and enhance Android AI interoperability, aiming to 'rebalance the playing field' and reduce Google's monopolistic hold on search and smartphone operating systems. Google has expressed concerns over user privacy and security, arguing the decision could expose private searches to unfamiliar companies. The move is part of the EU's efforts to promote competition and fairness in the tech industry.

Google DeepMind and Isomorphic Labs have launched a bioresilience program to prevent AI misuse in biology while aiding outbreak response, with a focus on preventing misuse, detecting outbreaks, and responding to attacks. The program has built over 15 partnerships with government bodies, biosecurity organizations, and research groups. The initiative aims to balance enabling scientific advances with keeping AI tools out of the wrong hands.

AI vendors are passing on their infrastructure costs to customers through increased prices and usage charges, with software budgets expected to rise as a result. Forrester warns that customers should prepare for bigger software bills next year, with 80% of decision-makers expecting data and software spending to increase. This shift is driven by the high costs of building and maintaining AI datacenters, estimated to reach $2 trillion by 2030.

NVIDIA has released Nemotron 3 Embed, a collection of open and commercially available embedding models that improve retrieval quality and provide practical deployment options for production-scale applications. The flagship model, Nemotron-3-Embed-8B-BF16, ranks #1 on the RTEB leaderboard. The models offer a range of features, including open weights, datasets, and recipes, and support multilingual and code retrieval.

Google Vids has introduced two new updates, Gemini Omni and personal avatars, to make video creation easier. Gemini Omni allows users to generate and edit high-quality videos using simple text prompts and image references, while personal avatars enable users to create a digital version of themselves to star in videos. These updates aim to simplify the video creation process and provide more personalized options for users.

Google's AI Mode now allows users to link and interact with select apps, including Instacart, Canva, and YouTube, enabling them to complete tasks across apps. This update expands AI Mode beyond answering questions and into task completion. The rollout aims to enhance user experience and compete with rival AI models.

Andrew Dai, a former Google DeepMind researcher, raised $55 million in seed funding at a $300 million valuation for his startup Elorian, which aims to advance visual AI. Dai shares his experience and insights on fundraising, pitching technical products, and recruiting top AI talent. His story offers practical lessons for founders navigating the AI landscape.

Moonshot's upcoming Kimi 3 model is expected to close the gap with Anthropic's Opus 4.8, with a parameter count between 2 trillion and 3 trillion, and will be the largest open-weight AI model from China. The release comes as the industry debates the value of closed-source models and considers open-source alternatives. Moonshot is also raising fresh capital, valuating the company at $31.5 billion.

Apple's generative AI offering, Apple Intelligence, has been approved for launch in China after partnering with Alibaba and Baidu to integrate their AI models into Apple's operating systems. The deal marks a significant step for Apple's AI ambitions in a key market, where the company generated $20.5 billion in sales in the second quarter. Apple Intelligence features will now be available to Chinese users with integrated AI capabilities like text and image understanding and generation.

1Password has introduced a new browser integration for Claude, allowing the Anthropic chatbot to access stored security credentials for tasks like booking travel and managing online accounts. This is made possible by a 'zero-exposure security framework' that keeps credentials secure. Users can authorize Claude to complete tasks without manually inputting login credentials.

OpenAI has developed GPT-Red, an LLM super-hacker designed to test and improve the safety of its models against cyberattacks. Meanwhile, heat pump sales in the US have doubled over the past 15 years, outpacing natural-gas furnaces. These developments highlight advancements in AI safety and the growing adoption of efficient heating technologies.

The European Union has ordered Google to give rival AI assistants and search engines greater access to Android and Google Search, potentially weakening Google's control over these platforms. The company must comply with the rules by January 2027 for search data and July 2027 for Android changes. This move could have far-reaching consequences for Google and its AI tool Gemini.

DharmaOCR, a specialized optical character recognition model, has outperformed newer models like Mistral OCR4 and Unlimited-OCR in Brazilian Portuguese due to its domain specialization and targeted training. The model's training pipeline involved supervised fine-tuning and Direct Preference Optimization, resulting in higher extraction quality and stability. This achievement highlights the importance of specialization in AI models.

The International Association of Chiefs of Police Technology Conference showcased AI's potential to automate routine policing tasks, which could significantly impact the legal process. Attendees were introduced to AI-powered solutions that may revolutionize the future of policing in America. The integration of AI in policing raises important questions about its implications.

Meta faced backlash for enabling an AI feature on Instagram by default, prompting users to opt out, and later rolled it back after three days. The incident highlights the growing concern over companies pushing AI features on users without their consent. Experts argue that companies should prioritize user privacy and provide opt-in options instead of opt-out defaults.

Google DeepMind and Isomorphic Labs have announced a joint approach to bioresilience, aiming to prevent the misuse of AI models and harness their capabilities to build a more resilient world against infectious diseases and future outbreaks. The partnership has advanced over 15 collaborations with government bodies and research groups to detect and respond to biosecurity threats. By making AI models available to trusted partners, they aim to support progress in prevention, detection, and response to biosecurity risks.

Neko Health has raised $700 million to expand its AI-powered body scan services in the US, starting with a clinic in New York. The company's preventive screening service combines medical imaging, blood tests, and clinician review to screen for potential health conditions. The funding brings Neko's total disclosed funding to over $1 billion.

Applied Computing, a London-based startup, has raised $20 million to develop an AI model for the oil, gas, and petrochemical industry, aiming to help operators make data-driven decisions in real-time. The company's foundation model, Orbital, combines sensor readings, physics, and chemistry to predict facility states and flag anomalies. With partnerships with major industry players, Applied Computing aims to revolutionize the industry's data analysis and decision-making processes.

Hugging Face detected and responded to a security incident driven by an autonomous AI agent system, which gained unauthorized access to internal datasets and credentials. The company has taken measures to fix the vulnerability, eradicate the attacker's foothold, and improve detection and alerting. The incident highlights the emerging threat of AI-driven attacks and the importance of AI-powered defense.

New York Governor Kathy Hochul is utilizing AI to review and update the state's outdated laws and regulations, aiming to streamline the legal code. The initiative targets antiquated rules, such as a $25 fee for dog hunting and restrictions on pregnant people working late hours. This effort is expected to significantly reduce the time required for a thorough review.

Google has introduced a new feature that allows users to connect their favorite apps directly to AI Mode in Search, enabling them to perform tasks like grocery shopping and playlist creation without leaving the search results. This integration aims to streamline daily tasks and make planning projects easier. The feature is starting to roll out in the U.S. this week.

Google is rebranding its AI note-taking app NotebookLM to Gemini Notebook, with the app remaining standalone despite deeper integration with Gemini and Google Search. The app has been continuously updated with features to organize and make sense of notes. The rebranding reflects Google's efforts to unify its AI offerings.

DoorDash has introduced a limited beta of DoorDash CLI, a command-line tool that allows developers to order food directly from their AI agent. The tool, called 'dd-cli', enables users to search stores, find deals, and check out. It's available to US and Canadian macOS developers via a waitlist.

Alexandre LeBrun, CEO of AMI Labs, avoids labeling his AI as 'AGI' or 'superintelligence' due to lack of clear definitions, instead focusing on developing 'world models' that can predict and interact with the physical world. This technology has potential applications in robotics, manufacturing, and healthcare. LeBrun is scouting for partners in Asia to access real-world environments and train his AI.

Despite the end of a key tax credit, heat pump sales in the US have continued to rise, outpacing natural-gas furnaces by 32% in the first quarter of 2026. The appliances' efficiency and potential to decarbonize buildings have driven their growing popularity. Heat pump shipments have seen a gradual increase since January, defying expectations of a decline after the tax credit expired.

OpenAI has released a ChatGPT basketball as part of its 'Pause. Play. Prompt.' campaign, aiming to promote physical activity and creativity outside of screens. The $70 basketball is made of 100% rubber and is suitable for outdoor play. The unusual product has raised questions about its target customer and the company's product-market fit instincts.