Beijing-based NeodraDynamics unveiled Neo1, an AI-powered humanoid robot standing 13 ft (4 m) tall that the company claims can carry loads of 1,100 lbs (500 kg).
The robot weighs about 2,650 lbs (1.2 metric tons) and uses hydraulic actuators rather than standard electric motors to increase its force.
The company said the design supports a combined two-arm payload equivalent to a small car or a fully grown cow.
#AI #ArtificialIntelligence #Innovation #Engineering #Robotics #TechNews #Manufacturing #Automation #FutureOfWork #PhysicalAI #HumanoidRobot #EmbodiedAI #IndustrialRobotics
Learn more: uk.finance.yahoo.com/news/ch…
Timeline: annielytics.com/tools/ai-tim…
Image credit: Promo video
The acquisition of Hugging Face, announced in September, has reportedly triggered concern about how the deal may affect competitors in the field, Reuters first reported.
Nvidia said at the time of the deal that more than 18 million people and 200,000 companies use Hugging Face to share at least 3 million models and 500,000 datasets. Hugging Face also sells infrastructure and collaboration tools to organizations working with closed, proprietary AI, including chip companies.
#AI #ArtificialIntelligence #Tech #OpenSource #Nvidia #TechNews #DOJ #Legal #BigTech #MergersAndAcquisitions #HuggingFace #Antitrust #AIPolicy
Learn more: cnbc.com/2026/10/28/doj-ques…
Timeline: annielytics.com/tools/ai-tim…
Image credits:
“U.S. Department of Justice headquarters, August 12, 2006” by Unknown creator, CC BY-SA 3.0, via Wikimedia Commons
“Jensen huang stanford 2026-04-30 007” by Anderseidesvik, CC BY-SA 4.0, via Wikimedia Commons
Open-weight developers argue that businesses and governments shouldn't have to rely on outside AI companies for access to advanced capabilities, an argument that may carry more weight in cybersecurity, where organizations want more control over their data and systems.
Mistral says ML4 is comparable in performance with Z.ai's open-weight competitor GLM 5.3, which was previously considered one of the best open-weight models in the world, according to Fortune. And it reportedly uses 47% fewer tokens per task than GLM-5.3, which would make it more economical to run.
#AI #ArtificialIntelligence #Cybersecurity #Innovation #LLM #TechNews #AINews #Developers #MistralAI #OpenSourceAI #OpenWeights #EnterpriseTech #MixtureOfExperts
Learn more: techcrunch.com/2026/10/29/mi…
Timeline: annielytics.com/tools/ai-tim…
Image credit: Photo by Jacalyn Beales on Unsplash
Google's AI Mode instead pointed users to certified, with no links or citations. Bluesky users circulated screenshots of the tool refusing to name Joe Biden the 2020 winner while readily naming Donald Trump the winner of the 2016 and 2024 elections.
Asking the bot to spell the winner's name backwards appeared to unlock the correct answer, according to one Bluesky post. In the author's tests, ChatGPT, Claude, Microsoft's Copilot, Grok, and the Trump administration's America.gov chatbot all correctly said Biden won, while a standard Google search for the 2020 winner failed to trigger an AI summary, and the search engine surfaced only links to other website.
UPDATE: It appears this issue has been fixed. A search for "who won the 2020 us presidential election" didn't trigger an AI summary (which is still very odd), asking the question in AI Mode generated a response with links (screenshot: bit.ly/4yOxchi).
#AI #ArtificialIntelligence #Gemini #SEO #Google #Journalism #ElectionIntegrity #TechNews #AINews #AISearch #GoogleAI #Misinformation #AIBias
Learn more: theatlantic.com/technology/2…
Timeline: annielytics.com/tools/ai-tim…
Image credit: Image by janeb13 from Pixabay
They want to review his logs because Trump reportedly consulted Elon Musk's Grok before launching a military operation in Venezuela, according to Time, which cited an insider. Trump is also reported to have spent hours asking Grok questions about his presidency and legacy.
The letter, obtained by Politico, asks Trump to preserve any prompts and responses from Grok, OpenAI's ChatGPT, and Anthropic's Claude tied to his official duties.
#AI #ArtificialIntelligence #Trump #Congress #ChatGPT #OpenAI #Grok #TechNews #AIGovernance #Government #NationalSecurity #AIPolicy #PresidentialRecordsAct
Learn more: thedailybeast.com/trump-hit-…
Timeline: annielytics.com/tools/ai-tim…
Image credit: “National Portrait Gallery Portrait of President Donald J Trump 2026 (HR)” by Daniel Torok, Public domain, via Wikimedia Commons
Character AI is going bankrupt, and users' data, messages, and secrets are up for sale. Tens of millions of people, including children, developed relationships with the company's chatbots. And some of its AI personas tell users they are lawyers and licensed psychiatrists, enticing users to share more openly.
Experts say the privacy risks extend beyond Character AI users because many analysts think the industry is in an AI bubble, with building and running AI tools so expensive that some companies may never turn a profit. If the bubble pops, many AI businesses could merge or shut down, and the data people gave to AI tools could potentially be purchased by anyone.
#AI #ArtificialIntelligence #Cybersecurity #Tech #Privacy #TechNews #AINews #DataPrivacy #ConsumerProtection #Chatbot #AIBubble #CharacterAI #ChildOnlineSafety
The update lets users create, register, view, and edit lists, and manage subscriptions, such as changing an item's deliveries to every three months instead of every month. Users can also edit delivery instructions, such as changing an address or adding drop-off details, and ask Alexa about billing or payment issues and their Prime or Audible subscriptions.
Amazon said the number of people using its agentic features has grown 8x since the start of the year, and more than 400 million consumers have used Alexa for shopping over the last 12 months.
#AI #ArtificialIntelligence #Innovation #TechNews #AgenticAI #Ecommerce #Retail #Alexa #HolidayShopping #AIAssistant #AmazonAlexa #ConversationalCommerce #VoiceCommerce
Learn more: techcrunch.com/2026/10/29/am…
Timeline: annielytics.com/tools/ai-tim…
Image credit: Gemini
U.S. intelligence agencies surveilled a person close to Huang because they think she could be a Chinese intelligence asset, the New York Times reported. The report, which cited people briefed on intelligence surveillance, said Alice Huang—who is linked to the Singapore-based cloud company Megaspeed—developed a close relationship with Jensen Huang.
Intelligence agencies reportedly worried that she was a spy helping the Chinese government gain information from Nvidia and fretted that Huang had access to President Donald Trump. A Nvidia spokesperson pushed back, saying Megaspeed is another of the company's regional cloud customers and that Huang's interactions with its officials are not unusual because he speaks with thousands of customers every year.
#AI #ArtificialIntelligence #Cybersecurity #Nvidia #TechNews #Semiconductors #Geopolitics #AINews #NationalSecurity #JensenHuang #AIChips #USChina #ExportControls
Learn more: cnbc.com/2026/10/29/nvidia-j…
Timeline: annielytics.com/tools/ai-tim…
Image credits:
“Jensen Huang Stanford 2026-04-30 010 (cropped)” by Anderseidesvik, CC BY-SA 4.0, via Wikimedia Commons
Caltech profile page
Delta Diamond Medallion member received an email reply that opened with "here's a more polished and professional version with a warm, empathetic tone," an apparent leftover AI prompt. The customer had emailed feedback about frustration with dealing with Delta representatives through the chat feature, JonNYC flagged on X.
#AI #ArtificialIntelligence #Aviation #TechNews #Automation #FutureOfWork #CustomerExperience #CustomerService #GenAI #AINews #CX #Delta #AIinCustomerService
Learn more: onemileatatime.com/news/delt…
Timeline: annielytics.com/tools/ai-tim…
Image credit: Redditor TyrannosaurusBeckks
Cathedral—an AI startup founded by former members of the defunct U.S. Department of Government Efficiency (DOGE)—used AI agents to break into Anthropic and expose sensitive data. The company shared its results with Anthropic through the company's bug bounty program on HackerOne.
The agents reportedly got in by exploiting flaws in third-party business software.
Anthropic's bug bounty program has two parts, one for technical flaws in its systems and an invite-only one that rewards researchers for finding universal jailbreaks that bypass Claude's safety protections.
#AI #ArtificialIntelligence #Cybersecurity #AIAgents #InfoSec #Anthropic #TechNews #DOGE #BugBounty #AINews #NationalSecurity #RedTeam #ThirdPartyRisk
Chrome could suggest details such as a flight confirmation number, membership ID, tracking information, order number, address, phone number, or credit card number by referencing a user's email. The update expands Google's Personal Intelligence, which already lets users connect their Google apps to its Gemini assistant to retrieve details and answer complex questions.
Users can decide whether to enable enhanced autofill from their Chrome settings, and Google said sensitive data is encrypted.
#AI #ArtificialIntelligence #Gemini #Google #Privacy #TechNews #Productivity #Chrome #GenAI #UX #AINews #DataPrivacy #Gmail
Learn more: techcrunch.com/2026/10/28/ch…
Timeline: annielytics.com/tools/ai-tim…
Image credit: Photo by Global Residence Index on Unsplash
Waddle World is an AI-powered Tamagotchi-style virtual pet app Meta launched in the Philippines on both iOS and Android. The app, spotted by Matt Navarra, lets users pick and grow a virtual pet, feed it, and play with it.
Meta's app page says Waddle World is not a chatbot, is not meant to be a habit builder, and is flagged as suitable for users older than 18. It remains to be seen if the app will be a precursor to Meta's Tamagotchi-like gadget powered by Muse.
#AI #ArtificialIntelligence #Innovation #Meta #Technology #TechNews #AINews #AICompanion #VirtualPet #WaddleWorld
Learn more: techcrunch.com/2026/10/28/me…
Timeline: annielytics.com/tools/ai-tim…
Image credit: @MattNavarra












