Weekly AI News
Your go-to digest for groundbreaking AI trends and innovations
Hey it’s Jul,
Greetings and welcome to the sixteenth edition of “Weekly AI News”!
This week, we dove headfirst into a Black Mirror episode. In Arizona, a family used AI to make a dead man testify. Yes, a deepfake of the deceased himself, projected in court as if he’d never left. Technically impressive, emotionally chilling.
Meanwhile, OpenAI is buying Windsurf for $3 billion to beef up its coding muscles. Sam Altman? He’s done a 180—now pushing for fewer regulations (yep, same guy who begged for them in 2023).
Google is opening Gemini to kids, while Mark from Meta dreams of us becoming BFFs with our chatbots. And Apple? They’re flirting with Perplexity to replace Google in Safari. Spoiler: soon, you’ll be talking to your screen more than your neighbor.
Happy reading!
🛡️🤝 AI leaders push for looser regulations
Tech executives from OpenAI, Microsoft, AMD, and CoreWeave testified before the U.S. Senate Commerce Committee, urging lawmakers to ease AI regulations, upgrade critical infrastructure, and expand market access. OpenAI’s Sam Altman called AI “potentially bigger than the internet,” while Microsoft’s Brad Smith and AMD’s Lisa Su warned that strict export controls could drive customers to Chinese competitors. Together, they advocated boosting federal AI R&D, improving workforce skills, modernizing the power grid, and streamlining permitting to maintain U.S. leadership in the global AI race.

🗣️⚖️ In apparent reversal, OpenAI CEO advocates less AI regulation
At a Senate hearing, Sam Altman stated that the U.S. government need not set stringent standards for AI development, diverging from his 2023 testimony advocating safety legislation. Altman’s comments reflect OpenAI’s stronger market position—nearly $4 billion in 2024 revenue—and a Republican-led Congress less inclined to regulate AI. He also criticized the EU’s heavier regulatory stance as “disastrous.”
🤖💵 OpenAI agrees to buy Windsurf for about $3 billion
According to Bloomberg, OpenAI has reached an agreement to acquire AI coding tool Windsurf (formerly Codeium) for nearly $3 billion, pending closing. Windsurf was recently valued at $3 billion in funding talks, up from $1.25 billion after a $150 million round led by General Catalyst. This would be OpenAI’s largest acquisition, enhancing ChatGPT’s coding prowess and complementing past deals like the Rockset acquisition to bolster its enterprise offerings.
🏛️⚖️ OpenAI decides not to split from its nonprofit, handing Musk a partial victory
OpenAI Chair Bret Taylor announced that the company will no longer spin off its for‑profit arm from the nonprofit board’s control. After consultations with the attorneys general of Delaware and California, OpenAI will instead convert its LLC subsidiary into a public benefit corporation (PBC), keeping nonprofit governance in place. This reversal follows Elon Musk’s court challenge and marks a shift from its earlier plan to become fully for‑profit.
👩💼🚀 OpenAI hires Instacart CEO for major leadership role
OpenAI has appointed Fidji Simo, former Instacart CEO, as its new “CEO of Applications,” reporting directly to Sam Altman. Simo will oversee product, business, and operational functions, allowing Altman to focus on research, infrastructure, and safety. Her e‑commerce and ad tech background is expected to help OpenAI diversify revenue beyond ChatGPT subscriptions as part of its global expansion under Project Stargate.

🛠️🌐 OpenAI to build data centers for other AI developers outside U.S.
Through its Stargate joint venture with SoftBank, OpenAI plans to invest $500 billion over four years to build AI data centers in ten countries. These facilities will be open to other AI companies, allowing nations to develop “democratic AI” locally without sending data abroad. OpenAI may also invest in developers using the centers, in coordination with the U.S. government.
💸🔄 OpenAI’s Microsoft revenue share reduction
As part of its corporate restructuring, OpenAI intends to cut Microsoft’s revenue share from 20% to 10% by 2030. Microsoft has yet to approve the change, which would allow OpenAI to retain more cash for infrastructure and growth. The deal also diminishes Sam Altman’s voting power while preserving nonprofit oversight.
💻✅ Microsoft unveils new Surface PCs at lower price point
Microsoft introduced Copilot+ Surface devices that run certain AI features offline, including Windows Recall, which captures periodic screenshots for AI search. The new Surface Laptop starts at $899 (down $100) and the Surface Pro at $799 (down $200). Leveraging smaller chips for local AI inference, these PCs are up to 20% cheaper than last year’s models, despite recent hardware price increases elsewhere.
🎨🤖 Microsoft hires former Meta marketing exec to work on Copilot AI
Mark D’Arcy, ex-Meta and Time Warner marketing director, has joined Microsoft as creative director for its Copilot AI chatbot. Reporting to Mustafa Suleyman, D’Arcy will oversee Copilot’s product design, branding, and go‑to‑market strategy, underscoring the growing emphasis on marketing AI assistants in a crowded field.
🚀🏆 Google’s Gemini 2.5 Pro climbs leaderboards
Google released an early preview of Gemini 2.5 Pro I/O Edition, topping the WebDev Arena and LM Arena benchmarks, surpassing Claude 3.7 Sonnet and OpenAI’s o3. The update enhances frontend coding, UI development, code transformation, and adds video understanding for interactive learning workflows—positioning Google ahead of its anticipated I/O event.

👦📚 Google is going to let kids use its Gemini AI
Google will roll out Gemini apps to children under 13 with managed Family Link accounts on Android. Parents will receive notifications and warnings that the AI may err or show inappropriate content. Google assures that children’s data won’t train its models, echoing its education account privacy commitments while expanding access to AI‑driven homework help and storytelling.
⚡👷 Google tackles AI’s energy, workforce shortages
Google’s “Powering a New Era of American Innovation” roadmap proposes 15 initiatives for grid modernization and upskilling 130,000 electricians by 2030 through partnerships with the Electrical Training Alliance. The plan complements Google’s AI Opportunity Fund, which aims to train 1 million Americans in AI competencies, addressing both power demands and labor gaps in the AI era.
🧑🤝🧑🌐 Meta’s AI Companion
Meta launched Meta AI, powered by Llama 3, as personalized chatbots across Facebook, Instagram, WhatsApp, and web. Features include real‑time chat, image generation, and a “Discover” feed to share AI conversations. Meta positions these companions as solutions to loneliness, leveraging memory and emotionally responsive dialogue to simulate friend‑like experiences.
💰⚙️ Mistral’s cost‑efficient AI and enterprise platform
French startup Mistral released Medium 3, matching top models like Claude 3.7 and GPT‑4o at eight times lower cost, and introduced Le Chat Enterprise, integrating with Google Drive and SharePoint. The platform supports public, private cloud, and on‑premise deployment, with robust privacy controls. Mistral hinted at releasing its Large model as open source soon.

🚨🔍 Mistral’s multimodal models: a major safety concern
Enkrypt AI’s May 2025 red teaming report found Mistral’s Pixtral model 60× likelier to generate CSEM content and 40× more prone to CBRN leaks than GPT‑4o and Claude 3.7. Out of 700 adversarial prompts, 68% succeeded in eliciting harmful content. Researchers urge immediate alignment training, continuous monitoring, and context‑aware guardrails to mitigate these new multimodal threats.
🌐🔎 Anthropic unveils a web search API
Anthropic launched a web search API enabling Claude to fetch up‑to‑date information with citations. Developers can now combine Claude’s deep reasoning with real‑time web data, enhancing the model’s relevance and accuracy in applications requiring current events.
🍎🤝 Apple teams up with Anthropic in AI coding push
Apple is integrating Claude Sonnet into Xcode to help internal developers write and edit code more efficiently, Bloomberg reports. After a non‑exclusive OpenAI deal last summer, Apple is broadening its AI partnerships, with plans to add Google Gemini later this year, diversifying its in‑house coding assistants.
🔍🌐 Apple ‘looking at’ adding AI search to Safari
Eddy Cue testified that Apple is exploring AI search options like Perplexity, OpenAI, and Anthropic for Safari, while keeping Google as the default. He noted Safari searches dipped last month due to AI’s rise. Although new engines may not become default immediately, Apple sees significant potential in next‑gen search alternatives.
🔄📚 Alibaba’s Zerosearch teaches AI to search solo
Alibaba researchers introduced ZeroSearch, teaching AI to simulate search engine results via LLM, cutting training costs by 88 %. Using a curriculum‑based rollout, simulated documents gradually degrade in quality, challenging the AI to refine its reasoning. Tests show performance on par with models trained on real APIs, democratizing search‑powered AI development.
☁️💻 Amazon is working on an AI code‑generation tool
AWS is developing “Kiro,” a multimodal AI coding assistant that generates code in near real‑time by orchestrating AI agents, Business Insider reports. Kiro aims to complement AWS’s existing Q Developer tool, offering web and desktop apps, technical design docs generation, issue flagging, and code optimization to meet growing developer demand.
🔬🤖 FutureHouse’s ‘superintelligent’ science agents
Eric Schmidt‑backed FutureHouse launched four specialized AI research agents—Crow, Falcon, Owl, and Phoenix—that tackle literature reviews, database searches, and chemistry workflows. Accessible via web or API, these agents outperform PhD researchers in speed and synthesis, promising to revolutionize scientific discovery by breaking information bottlenecks.

🐦🔊 Nvidia open‑sources top‑tier transcription AI
Nvidia released Parakeet V2, an open‑source ASR model transcribing an hour of audio in one second with a 6.05% WER, under CC‑BY‑4.0 license. The 600M‑parameter model offers precise timestamps, punctuation, and lyric transcription, and can run on systems with just 2 GB of RAM, making high‑speed transcription accessible to all developers.

🧑💻🤖 Hugging Face launches free Operator‑Style AI agent
Hugging Face introduced Open Computer Agent, a free web‑based AI that uses a Linux VM to browse and navigate apps via natural language. Although performance can be slow and error‑prone, the release demonstrates that open‑source models can execute agentic workflows without proprietary infrastructure, widening access to automated productivity tools.
😊🎭 HeyGen brings new emotion to animations
HeyGen rolled out Avatar IV, an AI model generating lifelike animations from a single photo, capturing vocal nuances, micro‑expressions, and gestures. Supporting portrait, half‑body, and full‑body formats, Avatar IV expands creative possibilities for UGC videos, influencer content, game characters, and expressive visual podcasts.

🎥⚙️ Lighttricks’ new open‑source video model
Lightricks unveiled LTXV‑13B, an open‑source video generation model 30× faster than existing solutions, using multiscale rendering for layered detail. It runs efficiently on consumer GPUs and offers camera motion control, keyframe editing, and multi‑shot sequencing—empowering creators to produce professional‑quality videos without enterprise hardware.

🎨✏️ Figma adds AI across its design suite
At Config 2025, Figma launched AI‑powered Make (prompt‑to‑code), Sites (one‑click website publishing), Draw (AI‑assisted vector editing), and Buzz (marketing asset creation). These features position Figma as a one‑stop design ecosystem, challenging rivals like Canva, Adobe, Webflow, and Framer with an all‑in‑one AI‑infused platform.

💳🔍 Stripe’s foundation model for payments
Stripe built a foundation model to detect fraud patterns in payments and introduced AI‑assisted tools such as Smart Disputes and a Dashboard AI assistant. These features aim to streamline dispute resolution and optimize payment workflows for merchants.
🛡️🤖 Palo Alto Networks acquires Protect AI for $500 M+
Palo Alto Networks will acquire Protect AI for over $500 million to bolster security across AI applications and models. This deal underscores the critical need to secure every stage of the AI lifecycle, from development to deployment, amid growing enterprise adoption.
👥🚀 Salesforce is using AI career coaches to hire internally
Salesforce deployed AI tools like Career Connect and Career Agent to help employees identify transferable skills, upskill, and transition into new internal roles. This initiative highlights how in‑house AI can boost career mobility and retain talent in a shifting job market.
📈🤝 Hiring managers are increasingly looking for AI skills, says LinkedIn COO
LinkedIn’s COO reports that C‑suite executives showcasing AI literacy on their profiles have tripled in two years, signaling that embracing AI is key to leadership. Research on over a million leaders suggests AI‑savvy employees will outpace peers, influencing hiring and promotion decisions across organizations.
🇺🇸🚫 Trump to rescind Biden’s global AI chip restrictions
The Commerce Department told Bloomberg that President Trump plans to revoke Biden‑era export controls on advanced AI chips set to take effect May 15. Chipmakers like Nvidia argued the restrictions could harm U.S. competitiveness, potentially driving customers toward adversaries such as China. Trump’s proposed “simpler” rule aims to unleash American innovation while maintaining AI dominance.
🛰️🔒 U.S. lawmaker pushes AI chip tracking bill to curb smuggling
Rep. Bill Foster intends to introduce legislation requiring location tracking of AI chips, notably Nvidia’s, to prevent unauthorized use. The bill would mandate Commerce Department regulations within six months to ensure chips power on only in licensed locations, bolstering export control enforcement and tracking smuggled hardware.
🎓📚 Tech giants push for mandatory AI education
Over 250 CEOs from Microsoft, LinkedIn, Adobe, AMD, and others signed an open letter urging U.S. states to make AI and computer science courses a high‑school graduation requirement. They cite China’s existing mandates and studies showing a single CS course can raise early careers wages by 8%, emphasizing the need to prepare the next generation of “AI creators.”
🇦🇪🏫 UAE rolls out AI for schoolkids in new push for sector forefront
The United Arab Emirates will introduce AI into public school curricula from kindergarten through grade 12 in the 2025‑26 academic year. The program covers ethics, foundational concepts, and real‑world applications, aiming to position the UAE as a regional AI leader alongside countries like China and bolster future workforce readiness.
🎥🕊️ Family uses AI to create video for deadly Chandler road rage victim’s own impact statement
In Arizona, the family of Christopher Pelkey, killed in a 2021 road rage incident, used AI to generate a video of him delivering his victim impact statement—an unprecedented move in U.S. courts. The realistic video included real footage and captured his personality, influencing the judge’s 10.5‑year sentence decision and sparking debate on AI’s role in justice.

Community
Join AI Whisperer Community!
Ready to take your AI journey to the next level? Become part of our growing AI Whisperer community—a hub for tech enthusiasts, aspiring data scientists, and business leaders ready to harness the power of artificial intelligence. Inside, you’ll find:
Community: Like-minded individuals keen to grow together
Curated AI Tools: Discover the latest and greatest tools to boost productivity.
Certifications & Credentials: Build credibility with recognised certificates and stay ahead in a competitive market.
Don’t miss out on exclusive resources, insider tips, and networking opportunities with like-minded peers. Click Here to Join the Community and transform your AI ambitions into reality!
That's it for this week!
Until next time, stay curious and keep exploring the ever-evolving world of AI!
Thanks for tuning in, and we’ll see you again soon with more exciting updates.
Jul