YouTube used its Made on YouTube event in New York City on September 23 to unveil a fresh batch of AI-powered creator tools. The updates, rolling out inside the Studio app, give creators AI feedback on video drafts, help with thumbnails and titles, and let them test up to three video cuts. YouTube also expanded shopping features to 35 countries and launched Custom Feeds for U.S. viewers. CEO Neal Mohan said on the Big Technology Podcast that AI will help creators grow bigger audiences without replacing the human connection that draws people to the platform.
Anthropic's Claude AI model helped researchers discover a previously unknown biological system called array-associated reverse transcriptases, or ART, which shares patterns resembling CRISPR gene-editing technology. Roughly 950 AI agents worked for 21 hours to scan more than 200,000 reverse transcriptases and narrow the field to 20 candidates. Anthropic launched a new life sciences research group and wet lab in the Bay Area for this effort. The company says Claude was the first to recognize the surrounding patterns and accessory protein, even though the underlying enzyme was already known.
In AI benchmarking news, GPT-6 Astra, OpenAI's newest flagship model, became the only system to complete a real driving course in a Toyota Corolla, navigating a cone course in a parking lot. The test, called DrivingBench, evaluated perception, planning, control, and latency. Other models tested included Claude Fable 5.1, Grok 4.6, and GPT-5.6 Sol. Meanwhile, Sky News reports that AI speed itself has become the biggest risk, with benchmarks jumping from 6% to 98% in a year in advanced math, and accuracy no longer tracking with reliability.
Google released two new text-to-speech models, Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS, offering over 2,000 production-ready voices across more than 100 languages. Flash TTS targets creative voice design and character voices, while Flash-Lite focuses on cost-efficient, high-volume production. Voice replication requires a 30-second audio sample and consent. Both models are available through the Gemini API and Google AI Studio. Separately, Qualcomm introduced the Snapdragon 8 Elite Extreme Gen 6 and Snapdragon 8 Elite Gen 6 mobile platforms, designed for next-generation flagship smartphones with improved AI, gaming, and camera features.
On the governance and security front, F5 showcased its AI Security Platform at events in Oman and announced an upcoming agentless tool called F5 Workforce AI Security. EY's Dilek Çilingir stressed that AI compliance requires understanding context and risks holistically, not just checking boxes. Hospices are increasingly using AI for clinical documentation, but experts from Husch Blackwell warn that shadow AI can lead to HIPAA violations and patient eligibility problems without clear policies and staff education.
Key Takeaways
- <ul><li>YouTube launched AI-powered creator tools at its Made on YouTube event, including draft feedback, thumbnail generation, and video cut testing inside the Studio app.</li><li>Neal Mohan stated AI will help creators grow audiences without replacing the human connection on the platform.</li><li>Anthropic's Claude AI discovered a novel enzyme system called ART that resembles CRISPR, with 950 AI agents scanning over 200,000 reverse transcriptases.</li><li>GPT-6 Astra from OpenAI was the only model to complete a real driving course in the DrivingBench test, outperforming Claude Fable 5.1, Grok 4.6, and GPT-5.6 Sol.</li><li>Sky News reports AI speed is now the biggest risk, with advanced math benchmarks jumping from 6% to 98% in one year.</li><li>Google released Gemini 3.8 Flash TTS and Flash-Lite TTS, offering over 2,000 voices across 100+ languages via the Gemini API.</li><li>Qualcomm unveiled the Snapdragon 8 Elite Extreme Gen 6 and Snapdragon 8 Elite Gen 6 for next-gen flagship smartphones.</li><li>F5 introduced its AI Security Platform in Oman and announced upcoming F5 Workforce AI Security, an agentless governance tool.</li><li>EY's Dilek Çilingir calls for holistic AI governance that considers data, algorithms, systems, and human oversight.</li><li>Hospices face growing HIPAA and compliance risks from shadow AI use without formal policies, according to Husch Blackwell experts.</li></ul>
YouTube shows off new AI tools and shopping updates
YouTube held its annual Made on YouTube event in New York City on September 23. The company revealed new AI-powered creator tools, shopping features, and personalized viewer options. Updates include AI feedback on video drafts, a Gemini-powered editing assistant, and expanded shopping in 35 countries. YouTube also introduced Custom Feeds and enhanced Ask YouTube features for U.S. users.
YouTube adds AI features to help creators make better videos
YouTube is rolling out new AI tools inside its Studio app to help creators improve their content. The draft-feedback feature uses AI to suggest changes to pacing, structure, and storytelling. Creators can now generate thumbnails and titles, test up to three video cuts, and track thumbnail performance automatically. YouTube also updated its analytics to explain why videos succeed and offer advice for improvement.
YouTube CEO Neal Mohan explains why he trusts AI
YouTube CEO Neal Mohan spoke on the Big Technology Podcast about the company's AI strategy. He described AI as a tool that improves the stage YouTube offers creators, including features for testing videos, tailoring thumbnails, and answering viewer questions. Mohan said he is not afraid of AI and believes it will help creators grow bigger audiences. He stressed that AI will not replace the human connection that draws viewers to the platform.
Claude AI helped find a new enzyme system similar to CRISPR
Anthropic announced that its Claude AI model helped discover a novel enzyme system called array-associated reverse transcriptases, or ART. Claude analyzed large DNA databases and identified a reverse transcriptase system with patterns resembling CRISPR gene-editing technology. This marks the first result from Anthropic's biology research efforts. The discovery was made when AI agents flagged an unusual repeating DNA sequence and an unknown protein next to the gene.
Anthropic's Claude helps make a biology discovery resembling CRISPR
Anthropic launched a new life sciences research group and wet lab in the Bay Area. Its Claude AI model discovered a previously unknown biological system with features similar to CRISPR. Roughly 950 AI agents worked for 21 hours to scan over 200,000 reverse transcriptases and narrow them down to 20 promising candidates. Anthropic says the ART system has three parts and appears mainly in bacteriophages. The company notes that while the underlying enzyme was known before, Claude was the first to recognize the surrounding patterns and accessory protein.
AI Infra Summit Day 2 features guest insights from Dustin Sklavos
Dustin Sklavos, a former AnandTech Senior Editor and Corsair product manager, wrote a guest post at the AI Infra Summit Day 2. He brings nearly two decades of industry experience, including managing hardware revenues over $100 million. The post covers threads and highlights from the summit event.
Sky News reports AI speed has become its biggest risk
Sky News reports that the speed of AI improvement is now its biggest risk. AI benchmarks jumped from 6% to 98% in a year in advanced math. The report notes that accuracy and reliability no longer track together in new AI systems. It also highlights the risk of AI building AI, which can speed up development beyond human understanding. The report lacks details on what safety measures labs are actually using.
F5 highlights AI security innovation at events in Oman
F5 held two events in Oman: F5 Academy Oman for customer training and a F5 Technology Showcase with Otech. The company showcased its F5 AI Security Platform for visibility and governance over enterprise AI. F5 also announced upcoming F5 Workforce AI Security, a new agentless tool. Sami Saber of F5 stated the platform helps organizations safely adopt AI in line with Oman Vision 2040.
EY's Dilek Çilingir discusses AI governance and compliance
Dilek Çilingir, a global AI and compliance leader at EY, shared insights on keeping AI governance in step with technology growth. She stressed that compliance requires understanding context and risks, not just checking boxes. Çilingir called for a holistic approach that considers data, algorithms, systems, and human oversight to manage AI risks proactively.
GPT-6 Astra becomes first model to drive a real car through a course
A new benchmark called DrivingBench tested AI models on driving a real Toyota Corolla through a cone course in a parking lot. GPT-6 Astra, OpenAI's newest flagship model, was the only one to complete the course. The test evaluated perception, planning, control, and latency in a real physical environment. Other models tested included Claude Fable 5.1, Grok 4.6, and GPT-5.6 Sol.
Hospices Face Growing AI Compliance Risks and Learning Gaps
Hospices are increasing their use of AI tools for clinical documentation and patient communication, but this creates legal and compliance risks. Experts from Husch Blackwell say staff using AI without official policies, called shadow AI, can lead to HIPAA violations and patient eligibility problems. Bryan Nowicki and Taylor Crossley warn that hospices need clear strategies, staff education, and acceptable use policies. They say early adoption helps hospices avoid falling behind as government and industry rules continue to develop.
Google Launches Gemini 3.8 Flash TTS and Flash-Lite TTS Voice Models
Google released two new text-to-speech models, Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS, as part of its Gemini Audio family. Flash TTS focuses on creative voice design and character voices, while Flash-Lite TTS targets high-volume, cost-efficient production. Developers can access over 2,000 production-ready voices and design new voices using prompts across more than 100 languages. Voice replication requires a 30-second audio sample and consent, and both models are available through the Gemini API and Google AI Studio.
Qualcomm Unveils Snapdragon 8 Elite Extreme Gen 6 and Gen 6 Mobile Platforms
Qualcomm introduced the Snapdragon 8 Elite Extreme Gen 6 and Snapdragon 8 Elite Gen 6 as two of the world's fastest mobile system-on-chips. The new platforms are designed to power next-generation flagship smartphones with improved AI, gaming, camera, and connectivity features. They provide a foundation for 5G, 6G, and future mobile technologies. Qualcomm says the chips deliver strong performance and power efficiency for leading global OEMs.
Sources
- YouTube unveils AI creator tools, shopping features at annual event
- YouTube releases new AI features for creators within its Studio app
- YouTube CEO Neal Mohan: Why We’re Betting On AI And Not Afraid Of It
- Anthropic says Claude AI helped discover novel enzyme system
- Anthropic Says Claude Has Helped Make a Biology Discovery, and It Looks a Bit Like CRISPR
- AI Infra Summit Day 2: Threads + Highlights
- Sky News says speed is now AI's biggest risk
- F5 puts the spotlight on AI security innovation in Oman
- In the AI governance race, EY's Dilek Çilingir explains how compliance can keep AI in step
- Driving Bench: GPT-6 Astra Becomes First Model To Drive A Real Car Through A Course
- Hospices Have ‘Learning Curves’ in AI Use, Risks
- Google Releases Gemini 3.8 Flash TTS and Flash-Lite TTS With Prompt-Based Voice Design
- Snapdragon Leads the Agentic AI Age with Two of the World’s Fastest Mobile SoCs Powering the Next Generation of Smartphones
Comments
Please log in to post a comment.