DeepSeek's New AI Model Scores 42.6 on HLE, Rivals OpenAI & Anthropic

DeepSeek released a new AI model called DeepSeek-V4.1-Flash-1017 on October 17. The company says it is faster, leaner, and cheaper to use than previous versions. It scored 42.6 on Humanity's Last Exam, up from 36.8, and improved on coding benchmarks. The release follows a five-hour service outage earlier in the week. DeepSeek competes with US firms like OpenAI and Anthropic, as well as Chinese rivals.

US consumers are growing uneasy about AI pricing tools used by retailers and restaurants. Electronic shelf labels are spreading at stores like Walmart and Whole Foods. Some social media posts falsely claim these labels scan faces or phones to set individual prices. Walmart CEO John Furner said the labels save time and are not used to charge different prices based on weather or individual shoppers. Lawmakers in at least 30 states have considered pricing bills this year.

The Federal Trade Commission proposed a policy warning businesses about failing to disclose how they use personal data for pricing. The agency said hiding personal data use in pricing could break consumer protection law. Consumer frustration continues to grow as dynamic pricing tools become more common in grocery stores and restaurants.

WhatsApp help pages accidentally exposed around 80 pages of hidden instructions meant for a Meta chatbot. Meta said a bug caused the pages to be public. The instructions included tips on what the chatbot should not say. Meta fixed the issue after the leak became public.

Circle and Proof announced a demo to help businesses trust AI agent transactions. The x401 protocol from Proof uses HTTP 401 to identify agents, while x402 uses HTTP 402 for payments. Together they let businesses pay identity providers and complete agent transactions using USDC.

An AI monitoring program exposed data belonging to Israeli political figures. The program scanned news sources every 90 seconds for mentions of Prime Minister Benjamin Netanyahu and his allies. Adviser Jonatan Urich left the code publicly accessible on GitHub, exposing private phone numbers and access tokens. Urich also connected his personal WhatsApp to Ultramsg, a third-party service from the United Arab Emirates.

nVent Electric, known as NVT, is building a third facility called Blaine 2 in Minnesota. It will expand liquid cooling capacity for AI data centers and is expected to open in the first half of 2027. NVT expects data center sales over $2 billion in 2026 and a backlog of $2.5 billion.

Firmable launched Tech Intelligence, a tool that shows B2B sales teams what software and technology a company uses. It monitors 10 billion internet endpoints weekly and matches data to about 20 million businesses. The tool is available as a paid add-on and through the Firmable MCP in Claude, ChatGPT, and Copilot.

India's Finance Minister Nirmala Sitharaman says India is now moving forward quickly in AI development. She notes the country started later than others but is catching up fast. She also stresses the need for proper training as industries change.

Director Tom Walsh made an AI generated Christmas movie called Merry the Christmas Witch. He spent $1.5 million on the film but says AI saved him tens of millions compared to traditional methods. He used ElevenLab for voices and Seedance and Nanobanana for video images.

Cohere has agreed to buy AISLE, an agentic AI cybersecurity company. AISLE found 29 new security flaws in curl and helped fix many other vulnerabilities. Cohere will invest in Prague as its global AI cybersecurity center and keep AISLE's founders in leadership roles.

Key Takeaways

  • DeepSeek released DeepSeek-V4.1-Flash-1017, scoring 42.6 on Humanity's Last Exam, competing with OpenAI and Anthropic.
  • US consumers are frustrated by AI pricing tools and digital price tags at Walmart and Whole Foods.
  • The FTC proposed a policy warning businesses about failing to disclose personal data use in pricing.
  • Lawmakers in at least 30 states have considered pricing bills this year.
  • WhatsApp help pages accidentally exposed around 80 pages of Meta chatbot instructions due to a bug.
  • Circle and Proof demoed protocols to help businesses trust AI agent transactions using USDC.
  • An AI monitoring program exposed private data of Israeli political figures after code was left public on GitHub.
  • NVT is building a third liquid cooling facility in Minnesota, expecting data center sales over $2 billion in 2026.
  • Firmable launched Tech Intelligence, monitoring 10 billion endpoints weekly, available in Claude, ChatGPT, and Copilot.
  • Cohere agreed to acquire AISLE, an AI cybersecurity company that found 29 security flaws in curl.

US consumers worry about AI pricing tools at retailers

U.S. consumers are growing uneasy as retailers and restaurants use AI and digital price tags to set everyday prices. Electronic shelf labels are spreading at stores like Walmart and Whole Foods. Some social media posts falsely claim these labels scan faces or phones to set individual prices. Lawmakers in at least 30 states have considered pricing bills this year. The Federal Trade Commission has proposed a policy warning businesses about failing to disclose how they use personal data for pricing.

Consumers angered by high-tech retail pricing methods

U.S. consumers feel confused and frustrated by technologies like dynamic pricing and AI tools that help set prices. Digital price tags are replacing paper labels in grocery stores. Walmart CEO John Furner said the labels save time and are not used to charge different prices based on weather or individual shoppers. Legislators in at least 30 states have considered pricing bills this year. The Federal Trade Commission proposed a policy warning that hiding personal data use in pricing could break consumer protection law.

Retailers adopt AI pricing tools amid consumer concern

U.S. consumers are angry about high-tech pricing methods used by restaurants and major retailers. AI tools help businesses set prices by predicting customer reactions. Digital price labels are spreading to stores like Walmart and Whole Foods. Walmart CEO John Furner said the company does not price people differently. Lawmakers in at least 30 states have considered pricing bills this year, and the Federal Trade Commission has proposed an enforcement policy about disclosing personal data use.

DeepSeek releases upgraded AI model V4.1-Flash-1017

Chinese AI company DeepSeek released a faster and leaner model called DeepSeek-V4.1-Flash-1017 on October 17. The new model scored 42.6 on Humanity's Last Exam, up from 36.8, and improved on coding and software development benchmarks. The update aims to reduce task time and lower usage costs. The release follows service disruptions, including a five-hour outage earlier in the week. DeepSeek competes with US firms like OpenAI and Anthropic as well as Chinese rivals.

China's DeepSeek debuts improved low-cost AI model

DeepSeek has launched an updated AI model that the company says outperforms American tools from OpenAI and Anthropic at a lower cost. The improved version follows the September release of the smaller V4.1-Flash model. Chinese AI developers are competing to deliver powerful systems at reduced prices. The release comes amid tech doubts and recent service slowdowns for DeepSeek. The updated model is available through DeepSeek's website, app and developer platform.

WhatsApp Help Pages Expose Meta Chatbot Instructions

WhatsApp's help pages leaked around 80 pages of hidden instructions meant for a Meta chatbot. Meta said a bug caused the pages to be public. The instructions included tips on what the chatbot should not say. Meta fixed the issue after the leak.

Circle and Proof Help Businesses Trust AI Agent Transactions

Circle and Proof announced a demo to help businesses trust AI agent transactions. The x401 protocol from Proof uses HTTP 401 to identify agents, while x402 uses HTTP 402 for payments. Together they let businesses pay identity providers and complete agent transactions using USDC.

AI Monitoring Exposes Israel Political Figures' Data

An AI program scanned news sources every 90 seconds for mentions of Prime Minister Benjamin Netanyahu and his allies. Adviser Jonatan Urich left the code publicly accessible on GitHub, exposing private phone numbers and access tokens. Urich also connected his personal WhatsApp to Ultramsg, a third-party service from the United Arab Emirates.

NVT Expands Liquid Cooling for AI Data Centers

nVent Electric, known as NVT, is building a third facility called Blaine 2 in Minnesota. It will expand liquid cooling capacity for AI data centers and is expected to open in the first half of 2027. NVT expects data center sales over $2 billion in 2026 and a backlog of $2.5 billion.

Firmable Launches Tech Intelligence for B2B Sales Teams

Firmable launched Tech Intelligence, a tool that shows B2B sales teams what software and technology a company uses. It monitors 10 billion internet endpoints weekly and matches data to about 20 million businesses. The tool is available as a paid add-on and through the Firmable MCP in Claude, ChatGPT, and Copilot.

India Finance Minister says country is no longer lagging in AI

India's Finance Minister Nirmala Sitharaman says India is now moving forward quickly in AI development. She notes the country started later than others but is catching up fast. She also stresses the need for proper training as industries change.

Director spends $1.5 million on AI generated Christmas movie

Director Tom Walsh made an AI generated Christmas movie called Merry the Christmas Witch. He spent $1.5 million on the film but says AI saved him tens of millions compared to traditional methods. He used ElevenLab for voices and Seedance and Nanobanana for video images.

Cohere to acquire AI cybersecurity company AISLE

Cohere has agreed to buy AISLE, an agentic AI cybersecurity company. AISLE found 29 new security flaws in curl and helped fix many other vulnerabilities. Cohere will invest in Prague as its global AI cybersecurity center and keep AISLE's founders in leadership roles.

Sources

NOTE:

This news brief was generated using AI technology (including, but not limited to, Google Gemini API, Llama, Grok, and Mistral) from aggregated news articles, with minimal to no human editing/review. It is provided for informational purposes only and may contain inaccuracies or biases. This is not financial, investment, or professional advice. If you have any questions or concerns, please verify all information with the linked original articles in the Sources section below.

AI Model Release AI Model Improvement AI Competition AI Pricing Tools Consumer Frustration Data Privacy AI Monitoring Data Exposure AI Cybersecurity AI in Business AI in Data Centers AI Tools for Sales AI in Film Production AI Acquisition AI Regulation AI in Politics AI in Finance AI in Retail AI in Cybersecurity

Comments

Loading...